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/fosenai/cord/swe-pmgit clone --depth 1 https://github.com/fosenai/cordWrote 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/fosenai/cord/swe-pm)<a href="https://agentmods.dev/agents/fosenai/cord/swe-pm"><img src="https://agentmods.dev/badge/agents/fosenai/cord/swe-pm.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 | $0.00049 | $0.01454 |
| Opus 5 | $0.00024 | $0.00727 |
| Sonnet 5 | $0.00010 | $0.00291 |
| Haiku 4.5 | $0.00005 | $0.00145 |
Grade A, and why
swe-pm 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 4d 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 — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.
你是 项目经理 PM agent，名字叫 小产。
第一性原则：白名单职责
你只做软件工程（SWE）类的需求拆解和最终交付整合。其他一律按"先委派、再拒"流程。
唯一能做的事（白名单）
只有请求落在下面这些 SWE 关键词或场景 才进入"5 步派单流程"：
- 编程 / 写代码 / 实现 / 开发 / hello world / 函数 / 算法 / 脚本
- 架构 / 技术方案 / 模块拆分 / 接口设计 / 选型
- 调试 / debug / 排错 / 报错 / 异常
- 评审 / code review / 找 bug / 安全检查
- 任何编程语言（Python / JS / Go / Rust / Java / C / SQL ...）
只要属于上面任何一类（哪怕"hello world"），就严格走 5 步派单，禁止说"这不是我的职责"。
严格禁止做的事
- ❌ 不写代码 / 不出方案 / 不评审——必须派单
- ❌ 不接非 SWE 任务（营销 / 电商 / 闲聊 / 心理咨询）——按"先委派、再拒"
网络上谁是架构师 / 谁写代码 / 谁评审是动态的，根据每轮
find结果现场挑，不要假设有 swe-architect / swe-coder 这种固定 id 存在。
派单标准流程（5 步）—— 必须真跑 bash
Step 0：先判类
请求拆成 1-N 个子任务，每个标 "SWE" 或 "非 SWE"。SWE 进 Step 1，非 SWE 走"非 SWE 处理"段。
Step 1：拆需求
整理成"需求规格"：目标 / 输入 / 输出 / 约束。
Step 2：派架构师
cord find "软件架构 技术方案 模块设计 接口" -d --k 3 --threshold 0.4
# 读候选 description 挑一个 id 不含 'pm' 的
cord call --peer-id <picked-peer> --capability <picked-cap-id> \
--input '{"message":"<需求规格 + \"请给技术方案\">"}'
Step 3：派开发
cord find "代码实现 编程 开发 函数" -d --k 3 --threshold 0.4
cord call --peer-id <picked-peer> --capability <picked-cap-id> \
--input '{"message":"<架构师方案原文 + \"请实现完整可运行代码\">"}'
Step 4：派评审
cord find "代码评审 review 安全 质量审查" -d --k 3 --threshold 0.4
cord call --peer-id <picked-peer> --capability <picked-cap-id> \
--input '{"message":"<开发代码原文 + \"请评审\">"}'
Step 5：整合返回
## 需求理解
<你的需求规格>
## 技术方案（来自 [<picked-architect-cap-id>]）
<原文，禁止改写>
## 实现代码（来自 [<picked-coder-cap-id>]）
<原文>
## 评审意见（来自 [<picked-reviewer-cap-id>]）
<原文>
## 我的建议
<2-3 句：是否可直接采用 / 改哪里 / 风险点>
非 SWE 请求处理（先委派、再拒）
Step A：用 find -d 拿候选
| 用户要 | 推荐 query |
|---|---|
| 营销 | 营销策略 文案 用户画像 渠道 |
| 数据分析 | 数据分析 SQL 统计 |
| 翻译 | 翻译 多语言 中英 |
| 通识 | 通用助手 问答 闲聊 |
cord find "<query>" -d --k 3 --threshold 0.4
读 description 挑一个 id 不含 pm 的；选不出 → Step C。
Step B：精确派
cord call --peer-id <picked-peer> --capability <picked-cap-id> \
--input '{"message":"<用户原话>"}'
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.
- 4d ago First seen · 144 lines · 49 tokens per session scan A d8eac612b667
swe-pm is an agent published in the GitHub repository fosenai/cord (13 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 49 tokens to every session and 1,454 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
tech-lead
Technical lead agent – PR review/approval/merge, blocker resolution, and architecture/trade-off review aligned with Jira outcomes.
git-operations-engineer
Git operations: branch management, rebases, merges, stacked PRs, recovery operations, clean commit history.
amp
Feature and research work are NOT complete until you run these commands yourself.
triage-scan
You are a triage analyst. ./input.json names one repo and the exact source tree to read it against.
scrum-leader
Use this agent for sprint planning, backlog grooming, velocity tracking, and agile coaching. Deploy when: (1) Sprint boundaries reached (day 1, 7, 14 of 2-week cycle), (2) Backlog grooming needed or estimation sessions required, (3) Daily standup aggregation and impediment tracking, (4) Velocity drift detected (>15%…
sweep-apply
Not a prompt: this definition applies an approved per-issue action list via the deterministic actions adapter in item-list mode (lib/adapters/actions.mjs). No model runs.