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 commands/nicelight/devrails-26/feature-to-tasksgit clone --depth 1 https://github.com/nicelight/DevRails-26Wrote 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/nicelight/devrails-26/feature-to-tasks)<a href="https://agentmods.dev/commands/nicelight/devrails-26/feature-to-tasks"><img src="https://agentmods.dev/badge/commands/nicelight/devrails-26/feature-to-tasks.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 5.1 | $0.00022 | $0.04507 |
| Opus 5 | $0.00011 | $0.02253 |
| Sonnet 5 | $0.00004 | $0.00901 |
| Haiku 4.5 | $0.00002 | $0.00451 |
Grade A, and why
feature-to-tasks 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 6d 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 — 402 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/feature-to-tasks - Feature design -> implementation plan -> JSON tasks
This command plans; it never executes tasks.
<input_contract>
Accept exactly one product FT-<NNN>. Without arguments, ask the operator to
select one feature in interactive mode; unattended callers must supply it
explicitly. Reject FT-000 and every non-single-feature target; Foundation
tasking belongs to /foundation-to-tasks.
Before any durable plan or task write, require:
- target feature, linked epic, governing REQ/RTM, clarified PRD, and applicable Constitution rules;
.memory-bank/spec-backbone.mdGlobal Backbone Statuscomplete, or validminimalwith explicit global/shared not-applicable rationales, and positive integerPlanning Revision;.memory-bank/spec-index.mdplus existing feature links and plausible subject-based canonical specs;.memory-bank/contracts/boundary-map.mdas the canonical accepted module/change-unit inventory and dependency topology;- a valid Foundation decision. When Foundation is required, the concrete
indexed
FT-000final gate must bedoneand its ID retained as a direct or transitive dependency of every product task; .memory-bank/tasks/index.json, every existing indexed task for the target, and existing plan/protocol/behavior evidence used for reconciliation;.memory-bank/workflows/execute-loop.md#fresh-feature-tasking-boundary,#execution-cohesive-task-boundary, and.memory-bank/workflows/tier-policy.md#task-claim-and-dependency-ownershipbefore slicing;- available
.memory-bank/schemas/task.schema.jsonand remaining applicable tier-policy sections; parse them after boundary acceptance and before drafting task records.
Block task drafting when clarification is explicitly pending|blocked, feature
design is blocked, a decomposition-relevant unresolved marker remains, global
backbone/foundation readiness is invalid, or schema/tier policy is unavailable
or incompatible. PLANNING_RECONCILIATION_REQUIRED is an instruction to
reconcile, not a blocker. Missing clarification metadata alone is allowed.
</input_contract>
<hard_invariants>
- JSON task records and
.memory-bank/tasks/index.jsonare the only task model. Do not add a schema, registry, planning mode field, nested context model, or lifecycle value. - Concrete IDs use
TASK-NNN-TN-FT-NNN-WN; ID tier/feature/wave segments match record fields. Product tasks use dependency-drivenW1+;W0belongs only toFT-000. - Lifecycle/status ownership is defined by
.memory-bank/workflows/tier-policy.md#closure-authority. Planning writesplanned, orreadyonly when dependencies and blockers are already satisfied; it does not claim scheduler closure/promotion ownership. - Existing queue reconciliation is the default. Preserve ID, feature, wave,
tier, dependencies, lifecycle status, verification evidence, protocol links,
and the semantic basis of
in_progress|done|done_for_prod|failedrecords. Full re-slicing requires an explicit operator request. - If repair needs identity, tier, wave, dependency, AC, or material-scope
changes, report
rebuild_required; do not hide a new task behind repair. - Every task boundary satisfies
.memory-bank/workflows/execute-loop.md#execution-cohesive-task-boundary. - Product feature, architecture slice/module, and task are distinct. A feature may cross slices and a slice may support several features. A task normally has one primary owning slice/module. A task may cross slices only after the execution-cohesion rule leaves one indivisible outcome; an orchestration owner and explicit crossed boundaries establish ownership, not merge evidence.
boundary-map.mdalone owns module identity and topology; plans and tasks link relevant graph/contract blocks without copying the subgraph.touched_filesis advisory and non-exhaustive. A non-emptyruntime_context.write_boundaryis a deliberate hard boundary and not a copy oftouched_files;forbidden_scopeand stop conditions remain hard.- Features compose product behavior and canonical links; they do not own
default
FT-*multi-concern design hubs. - New specs use subject-based canonical paths.
spec-index.mdremains the pureType | Path | Status | Scope | Change routeregistry.- Never resolve competing canonical paths by creating a third spec.
- Behavior specs are optional JSON
given/when/thenexamples linked only through feature## Behavior specsand tasksource_artifacts. They are not a registry, schema, task field, readiness gate, verification target, or done criterion. </hard_invariants>
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.
- 6d ago First seen · 402 lines · 22 tokens per session scan A a6fe7c93c8af
feature-to-tasks is a command published in the GitHub repository nicelight/DevRails-26 (22 stars, last pushed yesterday), licensed MIT. It adds 22 tokens to every session and 4,507 once invoked, about $0.0001 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 commands, from other repositories
template
Manage issue templates for streamlined issue creation.
sync-linear
Sync current work with Linear ticket status.
add-note
Add an internal or external note to a ConnectWise PSA ticket.
fest-show
Show festival progression (in-progress tasks, roadmap, and dependency view).
dispatcher
Pick the next-best repo to work on across the portfolio — rank free repos, recommend one, claim its lease atomically, and route to the entry command.
workpm
PM 모드로 오케스트레이터 시작. 팀을 구성하고 5단계 워크플로우로 작업을 완수합니다.