Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add SathiaAI/adversarial-review --skill intended-vs-implementedgit clone --depth 1 https://github.com/SathiaAI/adversarial-reviewWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sathiaai/adversarial-review/intended-vs-implemented)<a href="https://agentmods.dev/skills/sathiaai/adversarial-review/intended-vs-implemented"><img src="https://agentmods.dev/badge/skills/sathiaai/adversarial-review/intended-vs-implemented/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sathiaai/adversarial-review/intended-vs-implemented"><img src="https://agentmods.dev/badge/skills/sathiaai/adversarial-review/intended-vs-implemented.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00045 | $0.00593 |
| Opus 5.5 | $0.00018 | $0.00237 |
| Sonnet 5.5 | $0.00009 | $0.00119 |
| Haiku 4.5 | $0.00005 | $0.00059 |
Grade A, and why
intended-vs-implemented scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Intended vs. implemented
Documentation describes intent. Code describes reality. When they disagree, the code wins, and the disagreement is itself a finding worth surfacing.
Why this matters here
A roadmap, a handoff, a changelog, or a prior review report can confidently assert that a behavior shipped when it did not — a merge dropped it, a fix was described but implemented differently, or a status was aspirational. Building on the claim instead of the code propagates the error into your own work.
Worked example (real, this repo): the roadmap stated E3-S2b landed a
do_POSTslow-loris read-timeout. Checking the code onmain— nosettimeouton the POST read, no408, no handlertimeout— andgit log -S '408' -- scripts/mcp_server.py(empty) proved it never landed. Not a regression (nothing was lost); an overstated status. The fix was then correctly scoped into the story that actually owns it, and the discrepancy recorded.
The check
- Read the claim (roadmap status, changelog line, handoff, a "DONE" checkbox).
- Read the code that would implement it — the actual function, the actual test, the actual workflow. Not a summary of it; the lines.
- Prove presence or absence within a defined scope. Search the relevant source and generated
artifacts (
grep/pickaxegit log -S, run the test, inspect the output). An empty result is evidence only for the scope you searched — generated files, dynamic registration, aliases, and excluded paths can still hold the implementation — so record it as inconclusive when those remain possible, not as proof of global absence. - When they disagree, the code is authoritative. Update your plan to reality, and record the discrepancy (in the PR, the changelog, or a decision note) — do not silently "fix" the doc to match a reality you assumed, and do not silently build on the false claim.
Applies to your own output too
"[stated]" means someone said it, not that it is true. Your own tool output — a file's contents, a command's exit code — is what is, and it can contradict what you were told. Trust the primary source over the narrative every time, including when the narrative is yours.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 40 lines · 45 tokens per session scan A d36ea07fed6b
intended-vs-implemented is a skill published in the GitHub repository SathiaAI/adversarial-review (3 stars, last pushed 3d ago), licensed MIT. It adds 45 tokens to every session and 593 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-13.
Other skills, from other repositories
adversarial-reviewer
Stress-test a code change for concrete correctness defects, unsafe assumptions, and failure modes. Use when the user explicitly asks for an adversarial review, a deep bug hunt, or to tear a change apart, or to check your own change before submitting it. Do not use for ordinary review, style feedback, or general…
gsd-ns-review
quality gates | code review debug audit security eval ui.
gitnexus
A code-graph analysis add-on for examining an existing codebase, including symbols, call paths, execution flows, and effects across repositories. It can query GitNexus through its command-line or MCP interfaces.
issue
Use when starting a chain from a GitHub issue — turning an issue URL or number into a triaged, planned, dispatched, and reviewed pull request. Classifies the thread (bug → root-cause discipline, feature → plan chain, question → drafted reply), synthesizes a spec from the issue's own acceptance criteria, then runs the…
cleanup-code-inspections
Reduce technical debt and improve code quality by systematically resolving static analysis warnings.
superlint
This skill describes the mandatory standard operating procedure for using our internal SuperLint tool. Use this when tasks require fixing code quality issues according to corporate standards.