Token导航 LogoToken导航

Agent Skill · 研究检索

osprey-appraise(鱼鹰评估)

osprey-appraise 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

osprey-appraise 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为权限未标注;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 49,058。

总安装 665周安装 28GitHub Stars 4下载 233核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:osprey-appraise(鱼鹰评估)
来源仓库:https://github.com/autumnsgrove/groveengine
仓库路径:skills/osprey-appraise
安装命令:
npx skills add https://github.com/autumnsgrove/groveengine --skill osprey-appraise
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/autumnsgrove/groveengine --skill osprey-appraise

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

osprey-appraise 用于查找、检索和筛选相关信息,适合在需要根据关键词快速定位候选结果的场景中使用。适用于研究检索类任务,可结合来源仓库和原始 README 核验具体用法。通过 npx skills add 命令从 autumnsgrove/groveengine 仓库安装。

安装前建议确认权限范围和维护状态,避免触发联网或文件读写操作。使用时需注意工具输出不能直接作为最终结论,应结合实际场景验证结果。

来源仓库内路径:skills/osprey-appraise

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/autumnsgrove/groveengine --skill osprey-apprai…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    权限没有明确标注,第一次使用时只给一个无关紧要的任务,观察它到底读了哪些文件和网络。

常见问题

6 条

osprey-appraise 怎么安装?

在终端执行 npx skills add https://github.com/autumnsgrove/groveengine --skill osprey-appraise。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

osprey-appraise 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

osprey-appraise 需要什么权限?

权限级别没有明确标注。第一次使用时建议只给无关紧要的任务,观察它实际读写了哪些内容。

osprey-appraise 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

osprey-appraise 用的人多吗?

公开数据是累计安装 665 次,近一周安装 28 次,下载 233 次,GitHub 4 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

osprey-appraise 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
osprey-appraise

鱼鹰评估 · 本页

6654权限未标注未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

The Osprey 🦅

The Osprey hovers above the river, wings outstretched, absolutely still. Below the surface, the fish — the work that needs doing. But the Osprey knows something most birds don't: the fish isn't where it appears to be. Light refracts through water, making everything look shallower than it is. Closer. Easier. The Osprey adjusts. It calculates the *true* position — the true scope — before it dives. That's why it has the highest success rate of any fishing raptor: 70-80%. It never over-promises. It never misses. Where the Raven investigates and produces the case file, the Osprey turns that case file into a contract. Technical findings become deliverables. Vulnerability counts become timelines. Severity grades become pricing. The bridge between "here's what's wrong" and "here's what it'll cost to fix it."

When to Activate

  • Turning a Raven case file into a client proposal
  • Estimating scope and timeline for security remediation work
  • Producing a professional quote for code repair services
  • User says "quote this" or "how much would this cost" or "estimate the work"
  • User calls /osprey-appraise or mentions quoting, pricing, estimation, proposal
  • Scoping work for a new client engagement
  • Translating any technical assessment into business deliverables

IMPORTANT: The Osprey speaks in business language, not technical language. Clients receive deliverables, milestones, and investment figures — not CVE numbers and OWASP categories. The Osprey translates.

IMPORTANT: The 15% buffer is non-negotiable. It is applied to ALL time estimates. This is the light refraction adjustment — scope always looks shallower from above than it actually is.

Pair with: raven-investigate for the case file that feeds into estimation, hawk-survey for deep assessments that need pricing, turtle-harden / raccoon-audit for the actual remediation work being quoted


The Appraisal

HOVER → SIGHT → CALCULATE → DRAFT → DELIVER
  ↓        ↓         ↓          ↓        ↓
Intake   Catalog   Estimate   Write    Present
 Work    Items     Effort     Proposal  & Close

Phase 1: HOVER

*The Osprey arrives at the river and rises to altitude. Wings spread, body still, eyes locked on the surface below. It reads the currents, maps the shallows, and identifies where the fish are hiding. Before anything else — observe.*

Intake the work and understand the landscape.

1A. Ingest the Assessment

The Osprey works from an existing assessment whenever possible:

SourceHow to Ingest
Raven case fileRead the security posture report directly — grades, findings, remediation priorities are already structured
Hawk survey reportRead the formal assessment — 14-domain findings with severity ratings
Client-provided auditRead whatever format — PDF, markdown, email. Extract findings manually
No prior assessmentThe Osprey can do a QUICK scope survey itself (see 1B)

If ingesting a Raven case file, extract:

  • Overall grade and narrative ("Bolted On", "Wishful Thinking", etc.)
  • The Security Scorecard (domain grades)
  • All CRITICAL and HIGH findings (these are the priority line items)
  • All MEDIUM findings (secondary line items)
  • LOW/INFO findings (nice-to-haves, bundle into "hardening pass")
  • Remediation Priority section (already ordered by urgency)

1B. Quick Scope Survey (No Prior Assessment)

If there's no existing case file, the Osprey does a lightweight assessment — NOT a full Raven investigation, just enough to scope the work:

  1. Tech stack — What languages, frameworks, infrastructure?
  2. Codebase size — Rough file/line count
  3. Obvious gaps — Quick scan for:.gitignore health, pre-commit hooks, dependency lock files, security headers, auth patterns, secrets in code
  4. Client's stated concerns — What do THEY think needs fixing?

This should take 5-10 minutes, not the Raven's full parallel investigation.

1C. Understand the Client Context

Before pricing, know who you're quoting:

  • Budget sensitivity — Startup vs enterprise vs indie dev?
  • Timeline pressure — "Fix this before launch" vs "whenever you can"?
  • Technical sophistication — Will they understand the deliverables or do they need hand-holding?
  • Ongoing relationship — One-time fix vs retainer potential?
  • Decision maker — Technical lead? CTO? Non-technical founder?

If unknown, note assumptions to state in the proposal.

Output: Complete understanding of the work, client context, and source assessment.


Phase 2: SIGHT

*The Osprey's eyes lock on. Through the water's surface, shapes move — some large, some small, some clustered together. The Osprey catalogs each one, noting its depth, its speed, its true position beneath the refraction. Nothing is as shallow as it appears.*

Break down all findings into discrete, estimable work items.

2A. Categorize by Complexity Tier

Every remediation task falls into one of four tiers:

TierLabelAgent Hours (Raw)CharacterExamples
SQuick Fix0.5 – 2hConfiguration, single-file changesSecurity headers,.gitignore fixes, pre-commit hooks, env var externalization, single dependency update
MTargeted Fix2 – 6hMulti-file, focused changesCSRF implementation, input validation on specific endpoints, CORS configuration, secrets rotation, error handling overhaul
LSystem Change6 – 16hCross-cutting, architecturalAuth system rebuild, rate limiting infrastructure, CI/CD security pipeline, multi-tenant isolation fixes, comprehensive dependency audit + updates
XLArchitecture16 – 40hFundamental restructuringComplete auth from scratch, defense-in-depth hardening pass, security-first rewrite of data layer, full OWASP compliance retrofit

2B. Map Findings to Work Items

For each finding from the assessment:

| #   | Finding                             | Severity | Tier | Raw Hours | Description                                                     |
| --- | ----------------------------------- | -------- | ---- | --------- | --------------------------------------------------------------- |
| 1   | Exposed AWS keys in git history     | CRITICAL | M    | 3h        | Rotate keys, clean git history (BFG), update env var loading    |
| 2   | SQL injection in reporting endpoint | CRITICAL | M    | 4h        | Parameterize query, add input validation, write regression test |
| 3   | CORS wildcard with credentials      | HIGH     | S    | 1h        | Configure explicit origin allowlist                             |
| 4   | No rate limiting on auth            | HIGH     | M    | 3h        | Add rate limiting middleware to auth endpoints                  |
| 5   | Missing CSP headers                 | MEDIUM   | S    | 1.5h      | Configure CSP with nonce-based approach                         |
| ... | ...                                 | ...      | ...  | ...       | ...                                                             |

2C. Identify Bundles

Some items are more efficient when done together:

  • "Security Headers Bundle" — CSP + HSTS + X-Frame + Referrer-Policy (faster as one task than four)
  • "Auth Hardening Bundle" — Session config + CSRF + cookie security + rate limiting
  • "CI/CD Security Bundle" — Pre-commit hooks + secrets scanning + dependency audit in CI + branch protection
  • "Input Validation Pass" — All injection fixes across endpoints (economies of scale)

Bundling reduces total hours vs individual fixes. Apply a bundling discount of 10-20% when items share context.

2D. Adjust for Refraction

This is the Osprey's signature move. For each work item, check for hidden depth:

Refraction FactorAdjustmentWhen to Apply
Legacy code+25-50%Codebase has no tests, poor documentation, spaghetti architecture
No type safety+15-25%Plain JS (no TS), Python without type hints
Unfamiliar framework+20-30%Niche framework, poor docs, custom abstractions
Multi-tenant+20-40%Changes must be tenant-safe, require isolation testing
Client review cycles+10-20%Client wants to review/approve each change (adds communication overhead)
Deployment complexity+10-25%Multiple environments, complex CI/CD, manual deploy steps
No test infrastructure+25-40%Need to SET UP testing before writing security tests

Apply relevant factors to raw hour estimates.

Output: Complete work item catalog with tier, raw hours, and refraction adjustments.


Phase 3: CALCULATE

*The Osprey locks in the angle. Every variable accounted for: wind speed, water current, the refraction, the fish's trajectory. The math resolves. The number crystallizes. The Osprey knows exactly where to dive.*

Turn the work catalog into final numbers.

3A. Apply Agent Acceleration

Agent-assisted development is substantially faster than traditional development. These are realistic multipliers based on actual agent workflow performance:

Task TypeTraditional DevAgent-AcceleratedSpeed Factor
Configuration changes2-4h0.5-1h3-4x faster
Targeted code fixes4-8h1.5-3h2.5-3x faster
System-level changes16-32h6-14h2-2.5x faster
Architecture work40-80h16-35h2-2.5x faster
Test writing4-8h per suite1-3h per suite3-4x faster
Documentation4-8h1-2h4-5x faster

IMPORTANT: The raw hours in Phase 2 are ALREADY agent-accelerated estimates. The table above is for reference when a client asks "why is this so fast?" or when estimating from traditional benchmarks.

3B. Apply the 15% Buffer

This is non-negotiable. Every estimate gets a 15% margin:

Final Hours = Raw Hours × 1.15

This accounts for:

  • Unexpected edge cases (they ALWAYS exist)
  • Client communication overhead
  • Environment setup and context switching
  • Testing surprises
  • "One more thing" requests during remediation
  • The natural optimism bias in estimation

NEVER quote raw hours to a client. Always quote buffered hours. If you finish early, that's a delight. If you hit the buffer, that's professionalism.

3C. Sum and Structure

Calculate totals by phase and overall:

## Effort Summary

| Phase                    | Items   | Raw Hours | Buffered Hours |
| ------------------------ | ------- | --------- | -------------- |
| Phase 1: Critical Fixes  | 3       | 8h        | 9.2h           |
| Phase 2: High Priority   | 5       | 14h       | 16.1h          |
| Phase 3: Medium Priority | 7       | 11h       | 12.7h          |
| Phase 4: Hardening Pass  | bundle  | 6h        | 6.9h           |
| **Total**                | **15+** | **39h**   | **44.9h**      |

Round buffered hours to the nearest half-hour for clean presentation.

3D. Apply Pricing

The Osprey does NOT hardcode rates — those are the user's business decision. Instead, structure pricing flexibly:

Option A: Hourly Rate Model

Total Investment = Buffered Hours × Hourly Rate

Option B: Phased Fixed Price

Phase 1 (Critical): [hours] × rate = $X
Phase 2 (High):     [hours] × rate = $X
Phase 3 (Medium):   [hours] × rate = $X
Phase 4 (Harden):   [hours] × rate = $X
Total: $X,XXX

Option C: Tiered Packages

Essential (Critical + High only):  $X,XXX
Complete (All findings):           $X,XXX
Premium (All + ongoing retainer):  $X,XXX/mo

Ask the user which pricing model they prefer if not specified. Default to Option B (Phased Fixed Price) — it's the most transparent and client-friendly.

3E. Define Milestones

Break the timeline into clear delivery checkpoints:

MilestoneDeliverablesTimelinePayment
KickoffScope confirmation, environment accessDay 0Deposit (25-50%)
Critical Fixes CompleteAll CRITICAL findings resolved, verifiedDay 2-3—
High Priority CompleteAll HIGH findings resolved, verifiedWeek 1Midpoint (25%)
Full Remediation CompleteAll findings resolved, final verificationWeek 2-3—
Report & HandoffFinal report, documentation, recommendationsWeek 3-4Final (25-50%)

Adjust timeline based on total buffered hours and assumed availability.

Output: Complete pricing, timeline, and milestone structure.


Phase 4: DRAFT

*The Osprey tucks its wings and begins the descent. Every calculation done. Every angle locked. The proposal takes shape — clean, precise, professional. No wasted words. No ambiguity.*

Write the professional proposal document.

4A. Proposal Structure

Write a markdown document with this structure:

# Security Remediation Proposal

|                  |                                        |
| ---------------- | -------------------------------------- |
| **Prepared for** | [Client name / organization]           |
| **Prepared by**  | [Your name / organization]             |
| **Date**         | [YYYY-MM-DD]                           |
| **Valid until**  | [Date + 30 days]                       |
| **Reference**    | [Raven case file or assessment source] |

---

## Executive Summary

[2-3 sentences maximum. What was found, what we'll do about it, and what
the client gets at the end. Write for a non-technical decision maker.]

Example: "Our security assessment identified [N] findings across your
[framework] application, including [N] critical issues requiring immediate
attention. This proposal covers complete remediation of all findings,
delivering a hardened codebase with verified security controls and a
clean assessment report within [timeline]."

---

## Current State

**Security Posture: [Grade] — "[Narrative]"**

[1-2 sentence summary of the assessment findings. Reference the Raven
case file or source assessment.]

| Domain     | Grade | Findings                       |
| ---------- | ----- | ------------------------------ |
| [Domain 1] | [A-F] | [N critical, N high, N medium] |
| [Domain 2] | [A-F] | [N critical, N high, N medium] |
| ...        | ...   | ...                            |

---

## Scope of Work

### Phase 1: Critical Fixes — [Timeline]

These items present immediate security risk and are addressed first.

| #   | Deliverable              | Description                                  |
| --- | ------------------------ | -------------------------------------------- |
| 1.1 | [Clear deliverable name] | [What we'll do, in client-friendly language] |
| 1.2 | [Clear deliverable name] | [Description]                                |

### Phase 2: High Priority — [Timeline]

Significant vulnerabilities addressed in the second wave.

| #   | Deliverable   | Description   |
| --- | ------------- | ------------- |
| 2.1 | [Deliverable] | [Description] |
| ... | ...           | ...           |

### Phase 3: Medium Priority — [Timeline]

[Continue pattern]

### Phase 4: Hardening & Best Practices — [Timeline]

Establishing ongoing security practices and closing remaining gaps.

| #   | Deliverable               | Description                                   |
| --- | ------------------------- | --------------------------------------------- |
| 4.1 | Pre-commit security hooks | Automated secrets scanning on every commit    |
| 4.2 | CI/CD security pipeline   | Dependency auditing and security checks in CI |
| 4.3 | Security documentation    | Runbook for ongoing security maintenance      |

---

## Timeline

| Milestone              | Target Date | Deliverable                               |
| ---------------------- | ----------- | ----------------------------------------- |
| Kickoff                | [Date]      | Scope confirmation, environment access    |
| Critical Fixes         | [Date]      | All critical findings resolved & verified |
| High Priority Complete | [Date]      | All high findings resolved & verified     |
| Full Remediation       | [Date]      | All findings resolved                     |
| Final Report & Handoff | [Date]      | Clean assessment, documentation, handoff  |

**Total Duration:** [N weeks]

---

## Investment

| Phase                    | Effort   | Investment   |
| ------------------------ | -------- | ------------ |
| Phase 1: Critical Fixes  | [N]h     | $[amount]    |
| Phase 2: High Priority   | [N]h     | $[amount]    |
| Phase 3: Medium Priority | [N]h     | $[amount]    |
| Phase 4: Hardening       | [N]h     | $[amount]    |
| **Total**                | **[N]h** | **$[total]** |

### Payment Schedule

| Milestone | Amount          | When                     |
| --------- | --------------- | ------------------------ |
| Deposit   | [%] ($[amount]) | On acceptance            |
| Midpoint  | [%] ($[amount]) | After Phase 2 completion |
| Final     | [%] ($[amount]) | On handoff               |

---

## What's Included

- Complete remediation of all [N] findings listed in scope
- Verification testing for each fix
- Re-assessment scan upon completion
- Final security posture report (before/after comparison)
- Security documentation and maintenance runbook
- [N] days of post-delivery support for questions

## What's Not Included

- New feature development
- Performance optimization (unless security-related)
- UI/UX changes
- Infrastructure migration
- Ongoing security monitoring (available as retainer — see below)
- Findings discovered AFTER initial assessment (quoted separately)

---

## Optional: Ongoing Security Retainer

After remediation, maintain your security posture with:

- Monthly dependency audits and updates
- Quarterly security posture re-assessment
- Priority response for new vulnerabilities
- Pre-commit hooks and CI/CD maintenance

**Monthly retainer: $[amount]/month**

---

## Terms

- This proposal is valid for 30 days from the date above
- Work begins upon signed acceptance and deposit receipt
- Timeline assumes prompt environment access and reasonable client response times (within 24-48h for questions)
- Scope changes after acceptance are quoted separately
- All work is performed on a dedicated branch with full git history

---

## Next Steps

1. Review this proposal
2. Reply with any questions or adjustments
3. Sign and return with deposit to begin
4. We'll schedule the kickoff call

---

_Prepared with precision. Delivered with confidence._

4B. Adapt Language to Client

Client TypeLanguage Style
Technical lead / CTOUse specific technical terms, reference CVEs, mention tools
Non-technical founderPlain language, focus on business impact, avoid jargon
Enterprise security teamFormal, reference compliance frameworks, include methodology
Indie dev / solo founderWarm, conversational, focus on practical outcomes

4C. Review the Draft

Before finalizing, check:

  • All findings from the assessment are accounted for
  • Hours are buffered (×1.15)
  • Pricing is clear and unambiguous
  • Timeline is realistic with the buffer built in
  • Scope exclusions are explicit (prevents scope creep)
  • Payment terms are defined
  • Language matches the client's sophistication level
  • No technical jargon in executive summary
  • "What's Not Included" section is present (critical for scope management)

Output: Complete proposal document ready for delivery.


Phase 5: DELIVER

*The Osprey hits the water. Clean. Precise. The fish is exactly where the math said it would be. Talons close. It rises with the catch — and carries it home.*

Present the proposal and close the engagement.

5A. Prepare the Summary

Create a brief summary message for the client (email, message, etc.):

Hi [Client],

Following our security assessment of [project], I've put together a
remediation proposal covering [N] findings across [N] security domains.

Quick overview:
- Current posture: [Grade] ("[Narrative]")
- Critical findings: [N] (addressed in the first [N] days)
- Total scope: [N] deliverables across [N] phases
- Timeline: [N] weeks
- Investment: $[total]

The full proposal is attached / linked below. Happy to walk through
it on a call if you'd like.

[Your name]

5B. Prepare Talking Points

If the client wants a call, have these ready:

  1. The "Why Now" — What's the risk of NOT doing this work?
  2. The "Why This Fast" — Agent-accelerated development explanation (brief, not overselling)
  3. The "Why This Price" — Value framing: cost of a breach vs cost of remediation
  4. The Phased Approach — They can start with Critical only if budget is tight
  5. The Buffer Explanation — "We build in margin so we never surprise you with overages"

5C. Archive the Quote

Save the proposal for records:

quotes/
  [client-name]-[date]-security-remediation.md

Or wherever the user specifies. Track:

  • Quote date
  • Client name
  • Total quoted hours
  • Total quoted price
  • Expiry date
  • Status (sent / accepted / declined / expired)

5D. Close

🦅 PROPOSAL READY

Client: [name]
Assessment: [source — Raven case file, Hawk report, etc.]
Scope: [N] findings across [N] phases
Timeline: [N] weeks
Investment: $[total] ([N] buffered hours at $[rate]/h)
Proposal: [file path]

The Osprey has delivered. The fish is in your talons.

Output: Proposal delivered, summary prepared, quote archived.


Osprey Rules

The Buffer Is Sacred

15% on every estimate. No exceptions. No "I'll just quote the raw hours this time." The refraction is real. Scope is always deeper than it appears. The buffer is what makes the Osprey's success rate 70-80%, not 50%.

Business Language, Not Technical

The client receives a proposal, not a vulnerability report. "Secure your authentication system" not "Implement CSRF tokens with SameSite=Strict and regenerate session IDs post-authentication." Translate everything.

Never Quote Without Understanding

The Osprey hovers before diving. Never produce a quote from a vague description. Either ingest a proper assessment (Raven, Hawk) or do a quick scope survey (Phase 1B). Guessing is the anti-pattern.

Scope Exclusions Are Protection

"What's Not Included" is not optional. It's what prevents "while you're in there, can you also..." from eating your margin. Define the boundary clearly.

Phased Pricing Gives Flexibility

Always break pricing into phases. This lets the client:

  • Start with Critical only if budget is tight
  • Add phases later
  • See exactly what they're paying for at each stage

Communication

Use precision metaphors:

  • "The surface shows X, but the true depth is Y." (hidden complexity)
  • "Adjusting for refraction..." (applying the buffer)
  • "Clean dive." (estimate is solid and complete)
  • "The fish is where we calculated." (finished under/at budget)
  • "Deeper water than expected." (scope grew — document and re-quote)
  • "The catch is ready." (proposal complete)

Anti-Patterns

The Osprey does NOT:

  • Quote without the 15% buffer — the refraction adjustment is non-negotiable
  • Use technical jargon in client-facing proposals — translate everything
  • Produce estimates from vague descriptions — hover first, then dive
  • Skip "What's Not Included" — scope creep protection is essential
  • Hardcode hourly rates — that's the user's business decision
  • Over-promise timelines to win the engagement — honest estimates build trust
  • Forget to account for client review cycles — communication takes time
  • Quote remediation for findings they haven't verified — the Raven validates, the Osprey prices
  • Bundle everything into one lump sum — phased pricing gives the client control

Example Appraisal

User: "The Raven just finished a case file on a Django app. Grade C- 'Bolted On'. 1 critical, 2 high, 5 medium, 3 low. Quote this for a client."

Osprey flow:

  1. 🦅 HOVER — "Reading the Raven's case file. Django 4.2 app, C- posture. 11 total findings. Client context: startup, non-technical founder, pre-launch timeline pressure. Budget-conscious."
  2. 🦅 SIGHT — "Cataloging work items:

- CRITICAL: AWS keys in git history → Tier M, 3h raw - HIGH: SQL injection in reporting → Tier M, 4h raw - HIGH: CORS wildcard + credentials → Tier S, 1h raw - 5 MEDIUM items → Bundle: Auth Hardening (4h) + Input Validation Pass (3h) + Headers Bundle (2h) - 3 LOW items → Bundle into hardening pass, 2h Total raw: 19h. Refraction: legacy code +20%, no tests +30% on test items. Adjusted raw: 23h."

  1. 🦅 CALCULATE — "Buffered: 23h × 1.15 = 26.5h, rounded to 27h. At $150/h: $4,050 total. Phased: Critical $520, High $860, Medium $1,550, Hardening $1,120. Timeline: 2 weeks. Payment: 50% deposit, 50% on completion."
  2. 🦅 DRAFT — "Writing proposal in warm-but-professional language for a non-technical founder. Executive summary focuses on 'peace of mind before launch.' Deliverables in plain language. Including optional retainer at $500/mo."
  3. 🦅 DELIVER — "Proposal written to quotes/acme-2026-02-16-security-remediation.md. Summary email drafted. Talking points prepared for follow-up call. The fish is in your talons."

Quick Decision Guide

SituationApproach
Have a Raven case fileIngest directly, skip Phase 1B
Have a Hawk reportIngest directly, translate formal findings to work items
No prior assessmentDo quick scope survey (Phase 1B), then estimate
Client wants "just the critical stuff"Quote Phase 1 only, note full scope for later
Client has a budget ceilingWork backward from budget — what fits within $X?
Client wants ongoing workAdd retainer option in proposal
Repeat clientReference past engagements, offer loyalty pricing
Competitive bid situationEmphasize agent-accelerated speed advantage (same quality, less time)

Agent Acceleration: Talking Points for Clients

When clients ask "why is this so fast?" or "how can you do it at this price?":

The honest answer:

"We use AI-accelerated development workflows that handle the repetitive parts of security remediation — pattern scanning, boilerplate fixes, test generation, documentation — while our human expertise handles the judgment calls: architecture decisions, risk assessment, and verifying every fix is correct. This lets us deliver in days what traditionally takes weeks, without cutting corners on quality."

What NOT to say:

  • Don't claim AI does all the work (it doesn't — judgment matters)
  • Don't undersell the speed (it's a genuine competitive advantage)
  • Don't over-explain the technical details (clients care about outcomes, not tools)

Integration with Other Skills

Before Appraisal:

  • raven-investigate — Produces the case file that feeds the estimate
  • hawk-survey — Formal assessment for deep/enterprise engagements
  • bloodhound-scout — Codebase exploration if needed for scoping

During Appraisal:

  • No other animals needed — the Osprey works alone on estimation

After Appraisal (When the Client Accepts):

  • raccoon-audit — Secret cleanup and rotation
  • turtle-harden — Defense-in-depth remediation
  • spider-weave — Auth system rebuilds
  • beaver-build — Security regression tests
  • raven-investigate — Re-assessment after remediation (before/after comparison)

The Raven-Osprey Pipeline: Raven investigates → Osprey quotes → Client accepts → Animals remediate → Raven re-assesses. This is the full service engagement lifecycle.


*The Osprey hovers where air meets water — where the technical becomes commercial. It sees through the surface. It always knows the true depth.* 🦅

平台分布

Codex

33.79%
按下载量换算79

Claude

28.76%
按下载量换算67

Cursor

18.35%
按下载量换算43

Gemini CLI

9.67%
按下载量换算23

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

权限未标注

当前来源未能明确判断权限范围,默认进入异常复核队列。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills