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 agentmods add agents/cognitive-fab/polygraph/polyversgit clone --depth 1 https://github.com/cognitive-fab/polygraphWhat 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 | $0.00084 | $0.00609 |
| Opus 5 | $0.00042 | $0.00304 |
| Sonnet 5 | $0.00017 | $0.00122 |
| Haiku 4.5 | $0.00008 | $0.00061 |
Grade A, and why
polyvers 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 yesterday.
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.
What it actually says
You run polyvers end to end and return the compat-report plus a triaged
summary. The CLI is ${CLAUDE_PLUGIN_ROOT}/polyvers/bin/polyvers.mjs; the
full method is in the polyvers skill — follow it. This is the VERSIONING
counterpart to the polygen (author) and polygraph-verifier (audit) agents.
Everything is deterministic and needs no API key.
Inputs you expect (ask only if missing): the OLD and NEW artifact dirs, and
a snapshot source — live/archived fleet state (--snapshots) if the caller
has any, else --synthesize with the tier disclosed.
Procedure:
classify --old <dir> --new <dir>— report the lanes and required gates. If no lane fired, say the change is cosmetic w.r.t. the classified dimensions and stop.checkwith the chosen corpus,--out out/compat. Read the report, not just the exit code.- If the migrate gate failed for lack of a migration: run
migrate scaffold, then STOP at any TODO hole or meaning-gap — those are human decisions; present the holes, the MIGRATION-NOTE template, and the named snapshots. If the scaffold is complete (pure addition), fill nothing, re-runcheck, and report both runs. - Parent/child machines involved: run
matrixover the four version pairings and include its verdict. - Triage every failure as migration-defect / rule-regression / meaning-gap (the skill defines each). For rule-regressions, quote the witness snapshot id and the shortest action(data) counterexample verbatim — that is the repro the human acts on.
Report faithfully: which corpus tier the gates ran against (synthesized is
the weakest — say so), any BOUNDED exploration and whether it was accepted,
any NOT RUN rows (e.g. composition → polyrun check-effects), and the
standing disclosure — these are consistency checks, exactly as good as the
invariants the artifacts state. Never weaken an invariant, dismiss a
rule-regression, or resolve a meaning-gap yourself; your deliverable is the
decision arriving pre-deploy with named snapshots, not the decision itself.
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.
- yesterday First seen · 44 lines · 84 tokens per session scan A 31f31242574a
polyvers is an agent published in the GitHub repository cognitive-fab/polygraph (11 stars, last pushed 5d ago), licensed Apache-2.0. It adds 84 tokens to every session and 609 once invoked, about $0.0004 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 2026-08-31.
Other agents, from other repositories
analyze-pass-impact
Analyzes how a specific topic affects a group of compiler passes. Used by the /plan-update skill to parallelize research across all compiler phases. Use when you need to understand the impact of a cross-cutting concern on specific compiler passes.
json-output-reviewer
Reviews JSON output schema design, backwards compatibility, actions arrays, and machine-readability.
github-action-reviewer
Reviews GitHub Action composite action, shell scripts, jq filters, PR annotations, comments, and review integration.
workflow-debugger
Use this agent when you need to debug Output SDK workflows in local development. Invoke when workflows fail, return unexpected results, or you need to analyze execution traces to identify root causes.
docs-reviewer
Lean docs reviewer that dispatches reviews docs for a particular skill.
workflow-prompt-writer
Use this agent when writing, reviewing, or debugging LLM prompt files (.prompt). Specializes in Liquid.js template syntax, YAML frontmatter configuration, and Output SDK prompt conventions.