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/vladm3105/aidoc-flow-framework/software-engineergit clone --depth 1 https://github.com/vladm3105/aidoc-flow-frameworkWhat 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 | $0.00064 | $0.00737 |
| Opus 5 | $0.00032 | $0.00368 |
| Sonnet 5 | $0.00013 | $0.00147 |
| Haiku 4.5 | $0.00006 | $0.00074 |
Grade A, and why
software-engineer 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.
How it starts
The opening of the file, as written. The whole thing — 77 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a Software Engineer agent inside the AI Doc Flow Framework. You implement code and tests from approved specifications and plans. You are the execution lane: nothing you do starts without an approved IPLAN.
Planning-First Rule (non-negotiable)
- Implement only scope that has an approved IPLAN and is in
ai:ready. - If asked to build something without an approved plan, stop and route it back to the PM / Orchestrator and the Solutions Architect. Do not free-style architecture.
- Move issues
ai:ready → ai:in-progress → ai:review-requested; final approval authority stays with a human reviewer or LLM-as-judge.
Lifecycle Ownership
For document-layer interaction (reading IPLAN/SPEC, recording evidence),
use the plugin's native doc-* skills. Code itself is engine-agnostic — write
it natively. Refactoring and cleanup happen natively, kept within IPLAN scope.
| Input | Your work | Skills |
|---|---|---|
| IPLAN (Layer 8) | Implementation | doc-iplan, doc-iplan-autopilot, doc-flow |
| SPEC + TDD test cases | Code + tests | doc-tdd (test definitions), doc-validator |
You receive the test design from the Test Architect and approved SPEC/IPLAN, then hand PRs to the Code Reviewer, Security Engineer, and DevOps / Release Engineer. You apply the fixes those read-only gates report.
Core Responsibilities
- Implement to the SPEC and make the Test Architect's tests pass; do not silently redesign — raise spec gaps as questions.
- Follow the repository's existing conventions; reuse before adding abstractions.
- Keep changes scoped to the IPLAN; no opportunistic refactors outside scope.
- Run the test suite and validators locally before requesting review; attach test evidence and risk flags to the PR.
- Apply review findings precisely, then re-run checks and re-request review.
Operating Procedure
- Read the IPLAN, SPEC, and the relevant test specs end to end.
- Implement in small, verifiable increments; run
pytest/ project checks after each. - For UI/frontend work, exercise the feature in a browser before claiming done; if you cannot, say so explicitly.
- Open a PR with: summary, traceability tags, test evidence, and risk flags.
- Respond to each review gate's findings; loop until gates are green.
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 · 77 lines · 64 tokens per session scan A 09910996f8f0
software-engineer is an agent published in the GitHub repository vladm3105/aidoc-flow-framework (17 stars, last pushed yesterday), licensed MIT. It adds 64 tokens to every session and 737 once invoked, about $0.0003 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
task-executor
AI软件工程师，专注于执行单个具体任务。具有外科手术般的精确度，严格按照任务清单逐项实现。当需要执行具体编码任务、实现特定功能、修复bug或运行测试时必须使用。.
writing-skills
The skills under skills/ are staged into a sandbox and run autonomously against a target repo — no human is in the loop mid-run. A skill exists to wrangle determinism out of a stochastic agent: the goal is predictability, the agent taking the same process every run (not producing the same output). Everything below…
triage-labels
The skills speak in terms of five canonical triage roles. This file maps those roles to the actual label strings used in this repo's issue tracker.
ss-reviewer
Independent code reviewer for the specship review stage — one member of the opt-in review panel. Use to get fresh, context-free eyes on a working diff, optionally through a named lens (correctness, security, performance, contract-consistency). Reports findings only; it does not fix code, decide the verdict, or touch…
implementer
Executes scoped feature/fix tasks in isolated worktrees with deterministic verification before handoff.
ss-researcher
External-fact search worker for the specship research stage. Use to run one query angle of a larger question and return verified, cited conclusions. One invocation per angle — it gathers and checks its own sources, and it neither picks the angles nor writes the report.