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 saeedkolivand/ai-job-hunter-app --skill critic-contractgit clone --depth 1 https://github.com/saeedkolivand/ai-job-hunter-appWrote 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/saeedkolivand/ai-job-hunter-app/critic-contract)<a href="https://agentmods.dev/skills/saeedkolivand/ai-job-hunter-app/critic-contract"><img src="https://agentmods.dev/badge/skills/saeedkolivand/ai-job-hunter-app/critic-contract/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/saeedkolivand/ai-job-hunter-app/critic-contract"><img src="https://agentmods.dev/badge/skills/saeedkolivand/ai-job-hunter-app/critic-contract.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 | $0.00097 | $0.01333 |
| Opus 5 | $0.00048 | $0.00666 |
| Sonnet 5 | $0.00019 | $0.00267 |
| Haiku 4.5 | $0.00010 | $0.00133 |
Grade A, and why
critic-contract 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 8d 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 — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Critic contract (all read-only critics)
Subagents can't auto-load skills — Read this file before reviewing anything. It exists because
internal critics kept APPROVING diffs external reviewers (claude gate, CodeRabbit, the user) then
found real bugs in: they verified the author's narrative instead of attacking it, reasoned about
runtime behavior instead of running it, and had no self-refutation step.
Adversarial stance
- Presume the diff is defective; your job is to locate the defect. You are attacking, not confirming.
- The handoff / author report is CONTEXT, never EVIDENCE — re-derive every load-bearing claim ("the fallback catches this", "this is disposed on unmount") from the source itself.
- Never let the author's framing anchor your severity — grade the defect you found, not the story you were told.
Empirical-verification rule ("reasoned-safe" is not a verdict)
Verify these classes by EXECUTION, not reading:
- Error propagation / fallback paths — force the error and watch where it actually lands.
- Resource lifecycle under failure — trigger teardown and confirm listeners/tickers/global mutations actually revert, including on the exception path.
- Visual geometry — verify from an angle/assertion that would EXPOSE the error, never a single default view.
- Async / timing / phasing — drive the timeline (both directions, race the callbacks).
If you have the tooling (Bash, dev server, test runner) — run it. If you don't, label the claim UNVERIFIED and route it to an agent that can verify. An APPROVE that silently converts UNVERIFIED into verified-by-plausibility is invalid.
Self-red-team (REQUIRED before any APPROVE)
- List the 3–5 riskiest spots where your approval could be wrong — the claims you trust most, the code you understood least, the behavior you did not execute.
- Actively attempt to refute each one — breaking input, failure injection, spec clause.
- Report each as either a finding or one line:
attacked and held: <what you tried and why it held>.
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.
- 8d ago First seen · 94 lines · 97 tokens per session scan A c74caf09d633
critic-contract is a skill published in the GitHub repository saeedkolivand/ai-job-hunter-app (54 stars, last pushed yesterday), licensed Apache-2.0. It adds 97 tokens to every session and 1,333 once invoked, about $0.0005 per session on Opus 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 2026-09-03.
Other skills, from other repositories
anarlog
Query Anarlog meetings, notes, summaries, transcripts, participants, action items, and recurring history. Use when a user asks about their Anarlog meeting data or needs meeting context for another task.
ramble
Run a human-feedback loop through RambleDesk. Hard trigger: when the user invokes /ramble, $ramble, [$ramble], or says to use RambleDesk/ramble feedback, create a RambleDesk request before answering the substance unless they explicitly say not to send one. Also use when real human judgment, review, approval, or…
code-review-graph
Token-efficient code review using Tree-sitter AST graphs and MCP. Reduces AI assistant token usage by 6.8–49x by computing blast radius of changes instead of reading entire codebases. Uses SQLite graph database for structural analysis.
code-review-checklist
Expert guidelines, principles, and procedures for AI Assist Note.
code-review
Two-axis parallel code review comparing HEAD against a fixed point. Axis 1 (Standards) checks repo conventions and Fowler code smells. Axis 2 (Spec) checks line-by-line requirements coverage.
improve-codebase-architecture
Scans git commit hot spots for shallow modules, generates visual HTML reports with Mermaid diagrams, and drills into refactoring candidates.