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 skills add manishiitg/coding-agent-loop --skill agentworksgit clone --depth 1 https://github.com/manishiitg/coding-agent-loopWrote 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/skills/manishiitg/coding-agent-loop/agentworks)<a href="https://agentmods.dev/skills/manishiitg/coding-agent-loop/agentworks"><img src="https://agentmods.dev/badge/skills/manishiitg/coding-agent-loop/agentworks/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/manishiitg/coding-agent-loop/agentworks"><img src="https://agentmods.dev/badge/skills/manishiitg/coding-agent-loop/agentworks.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00064 | $0.01103 |
| Opus 5.5 | $0.00026 | $0.00441 |
| Sonnet 5.5 | $0.00013 | $0.00221 |
| Haiku 4.5 | $0.00006 | $0.00110 |
Grade A, and why
agentworks 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 19d 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 — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AgentWorks
This skill is an entry pointer, not a manual. All substantive guidance lives on the server and is fetched per task — nothing here can go stale.
This connection reads and runs, like the Slack and WhatsApp run-mode channels: tools read, and run-mode tools execute in pinned Run-mode sessions. Nothing creates, edits, or authors.
Connect
claude mcp add --transport http agentworks 'https://your-server/api/external/v1/mcp'
Approve the MCP connection in your browser. Its scopes allow reading (workflows:read, files:read) and running (runs:execute) workflows the account can access, reading (crews:read), asking or calling (crews:run), and creating and editing (crews:write) its Crews. The remote MCP surface has get_api_spec to discover available tool names and schemas, then call_tool to invoke one by name. Unavailable tools are omitted from the catalog.
First step
Use get_api_spec to inspect the available tools, then call get_agent_context through call_tool for your capabilities and guidance version. Discover workflow IDs with list_workflows first — IDs are never filesystem paths.
Guidance per task
List topics with list_guidance_topics and load only relevant ones via get_guidance_topic. Inspect workflow knowledge with list_workflow_knowledge / read_workflow_knowledge (learnings, knowledgebase notes, workspace skills, skill wiring). Use get_file_link for preview/download URLs.
Run
To run: call a run-mode tool such as execute_step — the reply carries session_id — then poll run_status for completion. Steer live work with send_step_message, stop it with stop_step / stop_all_executions, and read run evidence with list_runs, get_run, and get_logs. Schedules: list_schedules, get_schedule_runs, trigger_schedule.
Chat
chat asks the workflow assistant anything — analysis, explanations, follow-ups — in a pinned Run-mode session. Pass session_id to continue the conversation; sessions are shared with the run tools, so one conversation can ask, run, and ask about the run. Read replies with run_status, and answer waiting human-input steps with run_reply_input.
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.
- 19d ago Changed 1d99564984ff
- 21d ago Changed · -1 lines · -3 tokens per session 2485d8f7e8d7
- 22d ago Changed · +6 lines · +11 tokens per session 5fc0ed63ad60
- 23d ago Changed 4b52d8eab8b1
- 26d ago First seen · 40 lines · 56 tokens per session scan A b88f634f8d80
agentworks is a skill published in the GitHub repository manishiitg/coding-agent-loop (13 stars, last pushed 19d ago), licensed MIT. It adds 64 tokens to every session and 1,103 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-09-21.
Other skills, from other repositories
kungfu-agent-onboarding
Use when a user asks to understand, start, inspect, extend, or safely operate installed Kungfu; verify the installed pack, select one intent route, personalize the explanation, and propose one smallest safe next action.
machinist
Use Machinist to create, assign, and monitor software tasks. Use when a coding agent needs to work with Machinist, its GitHub issue workflow, intake labels, direct runs, or managed queue.
ship-it
Worktree-side implementation orchestrator for an OpenSpec change. Idempotent: gates automated scenarios on filesystem reality, owns the red-test fix loop, runs the docker harness with always-teardown, then drives ship-change inline. Escape hatch writes SHIPITBLOCKED.md. Runnable headless. Triggers: "ship it", "build…
goga-task-by-specifying
Interactive task formulation from a raw request.
goga-cell-go
Golang rules for implementing CODEMANIFEST contracts.
kano-backlog
Prioritize and refine a GitHub Issues backlog with the Kano model — categorize every open issue as Must-be, Performance, Attractive, Indifferent, or Reverse, apply Kano + priority labels back to GitHub automatically, and recommend the single best next issue to pick up. Use this whenever the user wants to triage…