A thin wrapper to execute and track progress of approved task plans located in the repository.
Install
mkdir -p .claude/skills/internal-gateway-execute-plans && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14907" && unzip -o skill.zip -d .claude/skills/internal-gateway-execute-plans && rm skill.zipInstalls to .claude/skills/internal-gateway-execute-plans
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Use when executing an approved compact or extended repository-owned plan from tmp/superpowers/<clear-action-or-task-name>/ and the numbered-file order, done-* loop, and blocker handling must stay explicit.Key capabilities
- →Execute approved retained plans from `tmp/superpowers/`
- →Resume interrupted retained plans
- →Apply repo-local closeout through status files
- →Confirm retained plan folder and approval to execute
- →Announce gateway and load `superpowers-executing-plans`
- →Require fresh passing task-level evidence before marking a task complete
How it works
The skill acts as a thin repository wrapper for approved retained-plan execution, owning repo-local start, task-transition, stop, and status-file policy, while delegating the execution loop to `superpowers-executing-plans`.
Inputs & outputs
When to use internal-gateway-execute-plans
- →Executing approved development plans
- →Resuming paused repository tasks
- →Closing out completed task plans
About this skill
Internal Gateway Execute Plans
Bundle References
references/execution-contract.md— repository hooks around the delegated execution loop.references/recovery-contract.md— continuation-first recovery and closeout decision ladder.references/status-contract.md— status transition table, required headings, and exact sibling filenames.scripts/plan_execution.py— read-only stdlib-only CLI for strict plan binding, structured recovery classification, status shape, resume safety, and completion readiness.
Referenced skills
/superpowers-executing-plansowns critical plan review, todo tracking, task execution, and its core stop behavior./internal-tddowns executable-behavior test-first guidance at the local task gate./superpowers-verification-before-completionowns final evidence before completion claims./addyosmani-code-simplificationis conditional and may be loaded only when the approved task explicitly authorizes simplification.
When to use
- Execute or resume an approved retained plan under
tmp/superpowers/plans/. - Apply repository-local preflight, task hooks, status handling, and closeout around the delegated core loop.
When not to use
- Writing, reformulating, reviewing, or challenging a plan.
- Running same-chat work that is not driven by an approved retained plan.
- Changing imported execution behavior or replacing the delegated core workflow.
Safety Boundary
The bundled CLI proves only mechanical safety: the plan is in the canonical retained directory, readable, actionable, and contains exactly one supported execution contract; status files are bound to the plan and fingerprint; and completion state is consistent. Missing required headings, execution fields, or contract data are blocking findings. Status files require the minimal resumable core plus closeout evidence for serialized routes. Conversational approval and runtime safety remain gateway responsibilities.
Gateway boundary
The gateway is a repository-owned extension of /superpowers-executing-plans. The
delegated owner supplies the core review, todo, execution, and stop loop. Keep
only these local responsibilities here:
- bind the exact plan path and explicit approval state;
- compute the SHA-256 fingerprint, run dirty-worktree preflight, and capture the plan-required validation baseline;
- apply task-level
/internal-tddand evidence hooks; - classify closeout evidence with
closeout-check, continue while a safe route exists, and preserve the baseline/final delta; - enforce the no-Git-mutation policy;
- replace the exact
DONE,PARTIAL,BLOCKED, orNEEDS_REVIEWsibling; - run resume and completion checks through
scripts/plan_execution.py.
Gateway phases
- Bind the approved retained plan, fingerprint, workspace overlap, and native validation commands. Completion: preflight passes and the baseline is recorded.
- Execute the delegated plan task-by-task with task-level red-first gates. Completion: each task has fresh focused evidence or a recorded safe pause.
- Recover through
references/recovery-contract.mdwhenever validation or execution is unresolved. Completion: the next candidate was tried, authority was requested when required, or exhaustion evidence is complete. - Decide with
closeout-check. Completion: continue immediately on acontinue-*orrequest-authorityroute, or write one legal status sibling for a terminal or explicit pause route. - Close with broader validation,
git diff --check, status binding, and the verification-before-completion gate. Completion: the status sibling and report contain the same fresh evidence.
Delegation checkpoints
Before loading /superpowers-executing-plans, bind the retained plan, record
approval, fingerprint the plan, capture the workspace baseline, and run the
plan's broad baseline validation. At each
task boundary, load /internal-tdd when the task changes executable or
evaluable behavior and require its red-first evidence before implementation.
After each delegated task, run the plan's focused validation, retain fresh
evidence, classify failures, and attempt bounded recovery while evidence
improves. Pre-existing or unrelated broad failures do not stop independent
tasks. Load /superpowers-verification-before-completion before any positive
completion claim; load /addyosmani-code-simplification only when explicitly
authorized by the plan.
Before a task transition or closeout, apply
references/recovery-contract.md. Preserve the native authoritative command,
continue on a safe continue-* route, and keep bounded search and retry evidence.
On pause or resume, preserve the plan fingerprint and use the status and resume
checks from scripts/plan_execution.py. At closeout, run the required broader
validation with the same commands used at baseline, record the baseline/final
delta, verify git diff --check, and write exactly one status sibling according
to references/status-contract.md. Always provide a concise user-facing report
with the outcome, changed work, validation, blocker or gap, recovery attempts,
and exact next action.
No-Commit Rule
Do not run git add, git commit, git push, git merge, or another Git
mutation while executing, pausing, or closing out a plan. Leave executed
changes uncommitted for the user to review. If a retained plan contains Git
mutation steps, skip them and record the plan drift in the status sibling.
Validation
git diff --checkpython3 scripts/plan_execution.py preflight <plan-file> --format compactpython3 scripts/plan_execution.py status-check <status-file> --format compactpython3 scripts/plan_execution.py resume-check <plan-file> <status-file> --format compactpython3 scripts/plan_execution.py closeout-check <plan-file> <evidence-file> --format compactpython3 scripts/plan_execution.py completion-check <plan-file> <status-file> --format compact- Confirm no live repository references point to removed bundle files.
The writer-owned versioned ## Execution Contract is authoritative for
validation IDs, native commands, required flags, equivalence policy, manual
obligations, and authority boundaries. The executor owns all six discovery
categories, recovery candidates, attempts, rejection evidence, authority
state, and closeout routing. request-authority keeps execution active and
does not produce a status sibling.
When not to use it
- →When writing, reformulating, reviewing, or challenging a plan
- →When running same-chat work not driven by an approved retained plan
- →When changing `superpowers-executing-plans` behavior
Limitations
- →Does not write, reformulate, review, or challenge plans
- →Does not change `superpowers-executing-plans` behavior
- →Does not create new closeout artifacts like `done-*` or `completion-report.md`
How it compares
This approach enforces strict execution discipline, including DRY, YAGNI, KISS principles, and tight feedback loops, ensuring that plan execution is precise and adheres to approved scope, unlike less structured task execution.
Compared to similar skills
internal-gateway-execute-plans side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| internal-gateway-execute-plans (this skill) | 0 | 1mo | Review | Advanced |
| github-project-management | 4 | 6mo | Review | Advanced |
| conductor-setup | 3 | 4mo | Review | Beginner |
| issue-manage | 2 | 5mo | Review | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by pagopa
View all by pagopa →You might also like
github-project-management
ruvnet
Comprehensive GitHub project management with swarm-coordinated issue tracking, project board automation, and sprint planning
conductor-setup
sickn33
Initialize project with Conductor artifacts (product definition, tech stack, workflow, style guides)
issue-manage
catlog22
Interactive issue management with menu-driven CRUD operations. Use when managing issues, viewing issue status, editing issue fields, performing bulk operations, or viewing issue history. Triggers on "manage issue", "list issues", "edit issue", "delete issue", "bulk update", "issue dashboard", "issue history", "completed issues".
gsd-executor
toonight
Executes GSD plans with atomic commits, deviation handling, checkpoint protocols, and state management
init-harness
mylukin
Creates AI agent task management structure with feature backlog (ai/tasks/), TDD enforcement, and progress tracking. Use when setting up agent-foreman, initializing feature-driven development, creating task backlog, or enabling TDD mode. Triggers on 'init harness', 'setup feature tracking', 'create feature backlog', 'enable strict TDD', 'initialize agent-foreman'.
team-coordination
alinaqi
Multi-person projects - shared state, todo claiming, handoffs