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.
git clone --depth 1 https://github.com/furkantokkan/agent-foundryWrote 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/commands/furkantokkan/agent-foundry/implement-task)<a href="https://agentmods.dev/commands/furkantokkan/agent-foundry/implement-task"><img src="https://agentmods.dev/badge/commands/furkantokkan/agent-foundry/implement-task/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/commands/furkantokkan/agent-foundry/implement-task"><img src="https://agentmods.dev/badge/commands/furkantokkan/agent-foundry/implement-task.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 | $0.00024 | $0.00585 |
| Opus 5.5 | $0.00010 | $0.00234 |
| Sonnet 5.5 | $0.00005 | $0.00117 |
| Haiku 4.5 | $0.00002 | $0.00059 |
Grade A, and why
implement-task 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.
What it actually says
Implement Task
Use the implement-task Codex skill for this request.
Treat the complete block below as one raw task input. Preserve quoted paths, logs, references, constraints, and natural-language flags; do not shell-tokenize it.
Resolve a recognized task ID only as production/tasks/<id>/contract.md; a
missing ID blocks and never falls back to an epic/story. Only non-ID input may
become a natural-language ad-hoc task.
An existing contract supplies persistent scope, ownership, acceptance, tests,
workflow, and risk. Command arguments may only apply invocation-local choices
or increase safety; a scope/ownership conflict must stop before mutation.
Default to preserving dirty work and making no commit.
Before mutation, detect post-implementation feedback for an existing tracked
task. Route it internally to task-bug [<ID>] <raw-feedback> instead of
starting a fresh implementation or deciding same-task scope. Only proceed
directly when this is initial execution, contractless ad-hoc work, or an
explicit CycleContext handoff from task-cycle. Automated passes never replace
required manual/visual evidence. Closed-task feedback follows the same route;
the downstream cycle may reopen it inside unchanged authority.
For a direct initial tracked task, require task state ready, attempt zero, an
empty defect ledger, and no CycleContext. For Unity, load unity-cli
first and unity-game-dev second before mutation. If independent verification
passes every required channel, compute the canonical reviewed/evidence
identities, write fresh acceptance evidence, set READY_TO_CLOSE, and return
the exact task-done <ID> next command without invoking an empty task-cycle.
Any populated/resumed defect lifecycle remains task-cycle-owned.
If a proven live owner prevents direct acquisition, do not fail the authorized
task as blocked. Queue it through agent-orchestration, wait for the
owner completion/lock-release signal, then revalidate and resume automatically.
Do not pre-announce a remembered defect ID, recurrence/reopen, root cause, or chosen code fix. Conversation IDs are candidates until task-cycle confirms the current canonical ledger and linked record.
Preserve all raw feedback so task-cycle can split multiple observable defects, deduplicate recurrences, update the contract's managed defect ledger, and continue until every same-task defect has fresh verification or a stop gate is reached. Do not create a replacement task or standalone bug file for those defects.
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 Changed 27444257389d
- 5d ago First seen · 54 lines · 24 tokens per session scan A 24fa7ae05dee
implement-task is a command published in the GitHub repository furkantokkan/agent-foundry (10 stars, last pushed today), licensed MIT. It adds 24 tokens to every session and 585 once invoked, about $0.0001 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 2026-09-25.
Other commands, from other repositories
checkpoint
Save verification state and progress checkpoint.
backlog
The project layer: milestones, epics, and user stories with refusing controllers — spec-seeded, sprint-ready.
tdd
Test-driven development with tests that actually catch breaks: red before green, name the break each test catches, and the mutation check before done.
test
Run the repository's actual test suite: every ecosystem's canonical runner — NOT run is never green.
status
The state of play, computed fresh: branch, dirty files, the active sprint, open work, index freshness.
validated
Record durable, git-tracked evidence that a ticket was actually validated (interactive-session twin of the governor's auto-promotion).