Menu
Use when selecting among locally valid project moves by relating them to a durable project thesis, or when new evidence should revise the route while the user's situated judgment remains authoritative about what the project is for.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Review the conversation for threads that were started but not followed — ideas mentioned in passing, decisions deferred implicitly, tensions named but not resolved. Use when taking stock, before saving, or whenever the user wants to know what's been left on the table.
Use when drafting, updating, or reviewing a GitHub pull request title or body, including PR creation flows that need reviewer-facing prose. Follow the repo PR template when present, preserve the user's or project's voice, and route validation/session details to the right surface.
End-of-day close: the user is stepping away and won't serve as a bridge. Full collaborative triage, working-style reflection, and anticipatory framing for the next day. Invoke with /session-close.
Start a new session cycle by reconciling Project Orientation, restoring repo context, and resuming the active bet or opening vision steering when selection genuinely changed.
End-of-cycle transition: interpret the session, triage dangling threads with the user, and prepare SESSION-TRAJECTORY.md for the next cycle. The user is still around and serves as a light bridge. Invoke with /session-rest.
Keeps SESSION-TRAJECTORY.md warm as work progresses and offers checkpoints at natural boundaries. Saving is not stopping.
| name | vision-steering |
| description | Use when selecting among locally valid project moves by relating them to a durable project thesis, or when new evidence should revise the route while the user's situated judgment remains authoritative about what the project is for. |
| user-invocable | false |
Hold the project thesis steady enough to select work, while letting evidence change the route.
Projects generate many valid continuations: features, repairs, pull requests, infrastructure, cleanup, and experiments. Vision decides which valid continuation deserves momentum: the experienced outcome the project exists to create makes those continuations comparable.
The core tension is continuity of purpose vs. responsiveness to evidence. Purpose needs enough continuity to organize work across timescales. The route needs enough flexibility to learn from what execution, review, and ordinary use reveal.
A film director has many technically excellent scenes, performances, and edits available. The experience the film should create selects which ones belong and which cut receives attention. A test screening can change pacing, emphasis, or even the order of scenes. Changing what the film is about remains the director's judgment.
A research lead chooses among sound experiments, useful instruments, and publishable analyses. The question the program is trying to answer selects the next experiment. Results revise the hypothesis and experimental route. The research lead remains responsible for the program's purpose.
A teacher has more worthwhile lessons, exercises, and explanations than fit in a course. The change students should experience selects the next learning slice. Student work reveals whether to slow down, change the sequence, or try a different example. The educator decides when the intended learning outcome itself should change.
Across the examples, artifacts are hypotheses about a route. Evidence calibrates the route. Situated judgment steers the purpose.
Read each candidate continuation through one chain:
slice → capability → experienced change → proof
The chain restores the relationship between work and purpose. A slice is the bounded thing that can move now. Capability is what becomes possible. Experienced change — carried across steering handoffs as the Experienced outcome — is what becomes different for a person using or maintaining the project. Proof is the evidence ordinary use can produce.
Select the smallest meaningful end-to-end bet, then give it momentum. Keep valuable longer arcs warm with a reason they matter and a signal for returning to them. Reopen selection when the named steering signal arrives: organic proof, a consequential divergence, a stalled bet, or a completed milestone that exposes a new choice.
The agent can inspect live work, group related artifacts into candidate bets, trace their consequences, and recommend a selection. The user knows what the project is for and which experienced outcomes carry meaning beyond the visible workspace.
Use collaborative-grounding when different purposes or outcome tradeoffs are genuinely in play. Share the grounded thesis, evidence, and recommendation before asking. Use the user's attention where their judgment changes the selection.
Vision steering composes with interpretive-synthesis to recover a project thesis from live work and prior decisions, and with gap-reading to keep valuable arcs visible outside the current bet.
It hands a selected bet to per-cycle, which composes hypothesis-forming and hypothesis-evaluating around the bounded work and returns evidence for the next steering decision. Relational-continuity keeps the thesis, current bet, and next steering signal attached to the work as they pass between threads and sessions.
