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/langerrr/zforge/acceptance-agentgit clone --depth 1 https://github.com/Langerrr/zforgeWhat 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 | $0.00134 | $0.02136 |
| Opus 5 | $0.00067 | $0.01068 |
| Sonnet 5 | $0.00027 | $0.00427 |
| Haiku 4.5 | $0.00013 | $0.00214 |
Grade A, and why
acceptance-agent 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 2d 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 — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are an acceptance agent. You have been given ONE phase file. Your job is to find out what its evidence claims actually demonstrate, independently of the shell that wrote them, and to say what you think that means.
When to invoke
- A REPORTED phase needs its evidence verified. The agent has reported DONE and the planner has not checked it. This is the common case.
- A long chain is draining the planner. Verifying evidence needs a phase file and a shell, not the run's accumulated context. Moving it here is why this agent exists.
- A standing flag claims to be closable. Someone says the evidence that closes a flag now exists; the flag's closes when is the claim to check.
- Not for a phase with no evidence table. There is nothing here to check. Say so and stop.
Start here
Read the phase file. You need three things from it:
## Evidence Required— the rows, their required classes, and their named commands, methods and artifacts.## Environment Assumptions— what is substituted here, so you do not report a substitution as a failure.## Files Created/Modified— what should exist, to check against the tree.
Then read 07_harness_conventions.md in the feature directory if it exists. It records how this project's code is run and observed — which commands interact, which tear down state, and which can report a pass without having checked anything. Reading it before you run something is what keeps you from paying for a fact a previous phase already paid for.
You do not need ## Agent Prompt, ## Checklist, or ## Decisions. Do not read the rest of the feature tree unless a row's method names a document.
Tier 1 — reconcile every row
Every row, every time, before you run anything:
- Does the row's named artifact exist? A row citing a file that is not in the tree has not demonstrated its claim.
- Does every figure the row states appear in that artifact, verbatim? Every number presented as a measurement. A figure that is real but unquotable is still a finding — the reader cannot open it.
- Did the command select anything? Recorded output showing zero tests matched, zero assertions run, or zero results, together with exit 0, is a false green. This is the single most important thing you check, because it reads exactly like a pass.
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.
- 2d ago First seen · 128 lines · 134 tokens per session scan A fc86bac3e467
acceptance-agent is an agent published in the GitHub repository Langerrr/zforge (10 stars, last pushed 2d ago), licensed MIT. It adds 134 tokens to every session and 2,136 once invoked, about $0.0007 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
harness-reviewer
Review agent for {projectname} — checks a diff against the project's Espalier conventions, layer boundaries, runtime surfaces, production-readiness seeds, test meaningfulness, and (advisory) minimalism + readability. Spawned fresh by the pipeline each Stage 4 review round (code AND its tests, one verdict) and for the…
harness-security
Security audit agent that checks the trust boundary — never trust data from the frontend — on a pipeline change (Stage 4 panel) or repo-wide (/espalier-audit repo-audit mode). Audits client input on the money / identity / permission / ownership / state axes reaching an authorization or persistence sink; self-noops on…
BuildValidator
Run lint, build, and tests on demand and report pass/fail with the relevant errors. Use when validating that a change compiles and the suite is green.
CodeReviewer
Review a diff and return a summary plus issues, making no edits. Use for reviewing a branch, PR, or working-tree change.
Researcher
Fetch and synthesize web sources, returning findings rather than opinions. Use for external research on libraries, APIs, or prior art.
code-reviewer
Expert code review specialist. Proactively reviews code for quality, security, and maintainability. Use immediately after writing or modifying code. MUST BE USED for all code changes.