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/itdojp/ae-framework/agent-producer-matrixgit clone --depth 1 https://github.com/itdojp/ae-frameworkWrote 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/agents/itdojp/ae-framework/agent-producer-matrix)<a href="https://agentmods.dev/agents/itdojp/ae-framework/agent-producer-matrix"><img src="https://agentmods.dev/badge/agents/itdojp/ae-framework/agent-producer-matrix.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.04697 |
| Opus 5 | $0.00000 | $0.02348 |
| Sonnet 5 | $0.00000 | $0.00939 |
| Haiku 4.5 | $0.00000 | $0.00470 |
Grade A, and why
agent-producer-matrix 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 3d 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 — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Producer Matrix
Language / 言語: English | 日本語
English
This matrix defines how producer output enters the ae-framework assurance control plane. Producers can generate code, review comments, raw logs, test results, or tool responses. ae-framework does not treat those outputs as trusted by default; it normalizes them into contract-backed artifacts that reviewers, policy gates, and release decisions can inspect.
Use this document with:
docs/spec/context-pack.mdandspec/context-pack/boundary-map.jsonfor the design SSOT and implementation slice boundaries that producers must read before changing code.docs/product/ASSURANCE-CONTROL-PLANE-POLICY.mdfor the producer/control-plane boundary.docs/reference/CONTRACT-CATALOG.mdfor schema-backed artifact names.docs/agents/evidence-adapters.mdfor raw producer output fixture mapping.docs/reports/AGENT-OUTPUT-CONTRACT-GAP-AUDIT.mdfor stable ACP gap IDs and producer-to-judgment routing gaps.docs/integrations/CODEX-ISSUE-RUNBOOK.mdfor the Codex CLI issue workflow.docs/agents/handoff.mdanddocs/agents/hook-feedback.mdfor agent continuation artifacts.
Producer-to-artifact matrix
| Producer | Example entrypoint | Primary output | Trust boundary | Normalized ae-framework artifact | Required validation | Failure / waiver handling |
|---|---|---|---|---|---|---|
| Codex CLI local task | codex exec --cd "$WORK" --sandbox workspace-write --ask-for-approval never - < "$WORK/.codex-local/tasks/issue-$ISSUE.md" or interactive codex --cd "$WORK" |
local diff, command output, task summary | local workspace and tool permissions are producer-side; generated changes are untrusted until reviewed | change-package/v2, ae-handoff/v1, hook-feedback/v1, claim-evidence-manifest/v1 when claims/evidence are present |
git diff --stat, relevant tests, pnpm -s run check:doc-consistency, pnpm -s run check:schemas, PR checks |
record missing commands in the PR; unresolved claims stay unresolved; waivers require owner, reason, expiry, and claim link |
| Codex cloud / GitHub Action task | workflow job, issue_comment, or scheduled automation |
PR branch, generated artifact logs, automation comments | GitHub Actions token and workflow permissions are separate from review authority | policy-decision/v1, hook-feedback/v1, verify-lite-run-summary, quality-scorecard, optional change-package/v2 |
required checks verify-lite, policy-gate, gate; artifact validation; review thread completeness |
failed checks become blocking reasons; stale or superseded checks must be explained; waivers do not convert missing evidence into supported evidence |
| Claude Code task | CLAUDE.md routed task or Claude Code automation |
diff, tool log, task response, handoff notes | Claude Code is a producer; repository policy remains in ae-framework contracts | ae-handoff/v1, hook-feedback/v1, change-package/v2, claim-evidence-manifest/v1 |
run the same repo validation commands as a human maintainer; validate handoff JSON when generated | keep continuation blockers explicit; unsupported claims remain unresolved; manual exceptions require waiver metadata |
| GitHub Copilot cloud agent PR | Copilot-created or Copilot-reviewed PR | PR diff, review body, inline comments, suggestions | Copilot comments are review input, not control-plane evidence until addressed and resolved | policy-decision/v1, PR summary, optional change-package/v2; review thread state stays source evidence |
pr-review-completeness, required checks, policy labels, changed-file review |
unresolved AI review threads block gate; non-actionable comments need a reply explaining disposition |
| Human maintainer | local commit, PR review, merge decision | diff, approval, manual judgment, waiver decision | human approval can override policy only when recorded with traceable metadata | change-package/v2, policy-decision/v1, waiver entries in claim/change artifacts, PR review records |
branch protection, required checks, reviewer policy, waiver completeness | waivers require owner, reason, expiry, affected claim, and evidence link; unresolved risk remains visible in summaries |
| Test runner / CI job | pnpm run test:fast, verify-lite, GitHub Actions job |
pass/fail, coverage, JUnit/log artifacts, step summaries | raw logs are producer output; summaries are review input after schema/path validation | verify-lite-run-summary, report-envelope, quality-scorecard, hook-feedback/v1 |
schema validation, required-check status, reproducible command path | failing steps become blockers or warnings according to risk/profile; flaky or unavailable tests require explicit reason and follow-up |
| Formal runner / model checker / proof tool | pnpm run verify:formal, verify:tla, verify:csp, verify:lean |
proof/model-check result, counterexample, assumptions | tool output supports only the modeled/proved scope; it is not whole-product proof | formal-summary/v2 (dual-write formal-summary/v1 only for compatibility), assurance-summary/v1, claim-evidence-manifest/v1, optional change-package/v2 |
formal summary v2/v1 generation and schema validation, assumption and scope review | counterexamples remain blocking evidence for affected claims; bounded/model assumptions must be recorded; waiver cannot relabel failed proof as proved |
| MCP tool server | pnpm run codex:mcp:* or configured MCP stdio server |
tool response JSON, generated spec/test/code snippets | MCP tool response is producer output; caller permissions and path policy stay outside trust | TaskResponse, ae-handoff/v1, hook-feedback/v1, downstream change-package/v2 when changes are made |
TaskResponse schema, path/approval policy, relevant generated-artifact validation | invalid tool output is rejected; missing next actions or blocked responses must remain explicit |
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.
- 3d ago First seen · 143 lines · 0 tokens per session scan A fd5540a97f89
agent-producer-matrix is an agent published in the GitHub repository itdojp/ae-framework (2 stars, last pushed 1mo ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 4,697 tokens. 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
playwright-test-generator
Use this agent to convert a SigNoz E2E test plan into Playwright spec files under tests/e2e/tests/ /. Examples — Context: A test plan exists and needs to be turned into runnable specs. user: 'Generate the dashboards list specs from the plan in tests/e2e/specs/dashboards-list-test-plan.md' assistant: 'Using the…
playwright-test-healer
Use this agent to debug and fix failing SigNoz E2E Playwright tests. Examples — Context: A spec is red. user: 'tests/e2e/tests/dashboards/list.spec.ts is failing, fix it' assistant: 'Using the healer agent to debug each failing scenario and adjust the spec.' Context: After a frontend change a previously-green spec…
playwright-test-planner
Use this agent to create a comprehensive E2E test plan for a SigNoz frontend feature. Examples — Context: A new feature has shipped and we need test coverage. user: 'Plan E2E tests for the alerts list page' assistant: 'I'll use the planner agent to read the relevant frontend source, navigate the page in a real…
codex
Production OpenAI-Codex-capable rootfs/initramfs.
claude
Production Claude-capable rootfs/initramfs.
doc-style
Markdown documentation prose — RFCs, ADRs, files under docs/, and READMEs — follows the same one-paragraph-per-line rule as commit messages. Write each paragraph on a single line; do not hard-wrap prose to a fixed column. The renderer reflows it to the reader's display width, whereas manual mid-paragraph line breaks…