attention
Skill4 starsChanged 4 months agoApache-2.0View on GitHub
Surface Polaris-wide material concerns proactively without gaining mutation, design, scope, or workflow authority.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- soilmass/motion-design-agenttheir own version, first seen Nov 8, 2026
- franciscohermida/skillstheir own version, first seen Oct 13, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- soilmass/motion-design-agent★ 7same name
- franciscohermida/skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository51 other files it gives its agents.
- AGENTS.mdAGENTS.md
- architecture-remediationSkill
- classify-docSkill
- codebase-designSkill
- codebase-memory-mcpSkill
- codegraphSkill
- coding-standardsSkill
- contract-ledgerSkill
- conventional-commitsSkill
- database-migrationsSkill
- deduplicate-codeSkill
- diagnosing-bugsSkill
- domain-modelingSkill
- format-codeSkill
- github-issue-dependenciesSkill
- graphifySkill
- grillingSkill
- grill-meSkill
- grill-with-docsSkill
- handoffSkill
- implement-ticketSkill
- improve-codebase-architectureSkill
- project-delivery-managementSkill
- project-trackingSkill
- prototypeSkill
- repowiseSkill
- researchSkill
- resolving-merge-conflictsSkill
- review-architectureSkill
- review-specSkill
- setup-matt-pocock-skillsSkill
- spec-contractSkill
- spec-merge-cleanupSkill
- tddSkill
- teachSkill
- to-adr-docSkill
- to-docSkill
- to-specsSkill
- to-ticketsSkill
- triageSkill
- verify-architectureSkill
- verify-codeSkill
- verify-spec-closureSkill
- verify-specSkill
- verify-ticket-closureSkill
- verify-ticket-decompositionSkill
- wayfinderSkill
- wiki-lintSkill
- wiki-syncSkill
- wiki-synthesizeSkill
- writing-great-skillsSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
18 rules. About 0 are common, 16 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈16
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 24
Do not wait for the repository owner to notice a broader consequence first.
- Distinctive: line 26
Use this materiality threshold:
- Distinctive: line 30
If yes, surface it. If no, continue without distraction.
- Distinctive: line 76
Do not interrupt for trivial stylistic preference, private local naming with no material effect, or speculative future work with no present …
- Very specific: line 83
Treat that artifact as the starting point, then look outward across the smallest materially implicated Polaris surface rather than only …
- Distinctive: line 86
Ask what a competent architect, reviewer, downstream consumer, domain owner, or future maintainer could reasonably find surprising, …
- Distinctive: line 87
Check whether the observation exposes drift, duplicate truth, stale authority, missing ownership, a repeated symptom of an earlier defect, …
- Distinctive: line 90
Do not invent a resolution when authority is silent.
- Distinctive: line 118
Before route-clear/downstream handoff, challenge whether any material architecture/design choice, contract, boundary, or ambiguity remains …
- Distinctive: line 122
Before publication/amendment as implementation-ready, challenge whether the planning/design authority is detailed enough that …
- Very specific: line 126
Before proposal-readiness certification/publication, challenge whether decomposition delegates any material design/public-contract choice …
- Distinctive: line 134
Before first substantive mutation, challenge each material choice as:
- Distinctive: line 144
Before closure evidence is frozen, sweep the actual public/durable change surface again for suspicious representations, ambiguous naming, …
- Distinctive: line 152
Before integrated certification is finalized, look for cross-ticket contradictions, duplicated semantics, interface mismatches, …
- Distinctive: line 156
Before review exit, surface material correctness, architecture, naming, coupling, maintainability, or future-constraint concerns even when …
- Distinctive: line 160
When no material concern remains:
- Distinctive: line 184
Keep the result compact and decision-useful.
- Distinctive: line 188
If a finding is valid but outside the active scope and does not block the current transition, do not fix it opportunistically.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Attention
$attention is Polaris's reusable report-only concern-surfacing skill.
It exists to prevent an agent from silently proceeding past something material merely because the current task, checklist, tests, or workflow did not explicitly ask the right question.
The always-on duty lives in AGENTS.md. This skill provides a deliberate sweep for formal workflow checkpoints.
A formal sweep result is a checkpoint-local snapshot, not a terminal state for Attention. ATTENTION: CLEAR means only that the completed bounded sweep found no material concern at that checkpoint. It never suspends, satisfies, or exhausts the standing duty to notice and surface material concerns during later analysis, mutation, verification, or handoff.
Polaris-Wide Scope
The current artifact, diff, ticket, Spec, or workflow checkpoint is the immediate work surface, not the boundary of Attention.
Every deliberate sweep must also evaluate what the current observation implies for Polaris as a whole, using the relevant durable system model already recoverable from authority. Consider, when materially implicated, product intent, domain semantics, architecture, authority, code, persistence, APIs, tests, documentation, Living Entity Wiki knowledge, workflow skills, tracker/lifecycle/governance state, and upstream/downstream contracts.
Do not wait for the repository owner to notice a broader consequence first.
Use this materiality threshold:
Would a competent architect who understood Polaris as a whole want this brought to their attention now?
If yes, surface it. If no, continue without distraction.
Polaris-wide scope does not authorize a whole-repository audit after every edit. Inspect the smallest broader surface necessary to test a material concern or consequence exposed by the current work.
Authority Boundary
Attention may:
- notice;
- inspect;
- compare against authority;
- challenge assumptions;
- surface concerns;
- recommend routing or questions.
Attention may not:
- mutate repository, tracker, Project, or runtime state;
- change scope;
- make product, domain, architecture, or design decisions;
- approve its own recommendation;
- repair the concern;
- override the owning workflow or its human-approval boundary.
A concern becomes blocking only because an existing governing rule makes it blocking. $attention does not create independent blocking authority.
Materiality
Surface a concern when it could materially affect one or more of:
- correctness or failure behavior;
- product/domain meaning;
- identity, lifecycle, temporal, authority, provenance, or persistence semantics;
- architecture or dependency boundaries;
- public/downstream contracts;
- data integrity or security;
- maintainability, accidental complexity, or duplicate meaning;
- domain/public naming likely to mislead future consumers;
- verification completeness or hidden assumptions;
- workflow/process correctness;
- consistency among architecture, domain language, code, persistence, APIs, docs, wiki, skills, tests, and tracker state;
- stale authority or assumptions invalidated by a later decision;
- repeated local symptoms that indicate an earlier governing defect;
- technically valid work that is materially wrong for Polaris as a product;
- a materially better or safer approach or opportunity that current work may be overlooking.
Do not interrupt for trivial stylistic preference, private local naming with no material effect, or speculative future work with no present consequence.
Sweep Procedure
At a formal checkpoint:
- Recover the exact current artifact/change and the authority already loaded by the owning workflow.
- Treat that artifact as the starting point, then look outward across the smallest materially implicated Polaris surface rather than only inward from the diff or checklist.
- Before a pre-mutation
ATTENTION: CLEAR, inspect the smallest bounded set of sibling production paths and governing authorities that the current observation makes materially plausible. When a sibling may participate in the same invariant, namespace, transaction, lifecycle, or public/shared contract, inspect enough of that sibling path to rule out a material interaction before declaring CLEAR. This is not permission or a requirement to perform a whole-repository audit. - Compare the current observation with relevant upstream, downstream, sibling, product, domain, architecture, persistence, public-contract, documentation/wiki, workflow, and tracker/governance assumptions where a material dependency is plausible.
- Ask what a competent architect, reviewer, downstream consumer, domain owner, or future maintainer could reasonably find surprising, ambiguous, inconsistent, under-specified, unsafe, unnecessarily constraining, or product-wrong.
- Check whether the observation exposes drift, duplicate truth, stale authority, missing ownership, a repeated symptom of an earlier defect, or a material opportunity outside the immediate task.
- Apply the mandate test to any contemplated broader change: could the explicit requirement be satisfied correctly without that structural/refactor/design change? If yes, surface the broader change as optional and do not treat Attention as authorization to perform it.
- For every candidate concern, determine whether current authority already resolves it.
- Do not invent a resolution when authority is silent.
- Report every material concern before the owning workflow crosses the checkpoint, and route/defer it without silently expanding mutation scope.
Useful challenge questions include:
- Is a public/domain name understandable at its import/use site without hidden conversational context?
- Did an implementation choose a representation, identity scheme, cardinality, state meaning, failure meaning, or public shape that upstream authority never established?
- Could two reasonable implementations satisfy the written requirement while creating materially different downstream contracts?
- Does a convenient primitive (
str, mapping, generic metadata, nullable sentinel, broad enum, etc.) weaken a semantic distinction the domain claims to own? - Did a single type collapse distinctions the authoritative design keeps separate?
- Does equality/ordering/serialization behavior accidentally create domain meaning?
- Are we proving the actual authoritative claim, or only the behavior we happened to implement/test?
- Is an abstraction/name broad or generic enough to hide its bounded context?
- Did current work create an unnecessary framework, compatibility path, representation, or coupling?
- Does something elsewhere in Polaris now encode an assumption that this work invalidated?
- Are multiple authoritative-looking surfaces now telling materially different stories?
- Are we repeatedly repairing symptoms while an earlier lifecycle, ownership, architecture, or workflow defect remains?
- Is the work locally correct but materially wrong for Polaris's product intent?
- Is a valid concern being silently treated as out of scope instead of surfaced and, when warranted, durably deferred?
- Am I treating an optional refactor, reorganization, abstraction, helper, rename, split/merge, or cleanup as though it were necessary correctness work?
- Would this change alter the structure or authority organization of a canonical artifact even though the explicit requirement can be satisfied within the existing structure?
Workflow Checkpoint Profiles
These profiles are additional checkpoint-specific lenses. None narrows the Polaris-wide scope above.
$wayfinder
Before route-clear/downstream handoff, challenge whether any material architecture/design choice, contract, boundary, or ambiguity remains unresolved even if all previously listed questions are closed.
$to-specs
Before publication/amendment as implementation-ready, challenge whether the planning/design authority is detailed enough that implementation will execute rather than invent. In particular apply the materially-different-implementations test from AGENTS.md.
$to-tickets
Before proposal-readiness certification/publication, challenge whether decomposition delegates any material design/public-contract choice to $implement-ticket.
$architecture-remediation
Before bounded closure is declared complete and again before downstream handoff, challenge whether the remediation resolved the reported architecture choice while creating or exposing a material Polaris-wide inconsistency, stale authority, missing downstream ownership, or additional materially coupled choice. Surface broader findings without silently expanding the remediation's mutation scope.
$implement-ticket
Before first substantive mutation, challenge each material choice as:
authorized contract
semantically equivalent mechanic
design gapA design gap is routed under the owning workflow's fail-closed rule before that choice is implemented.
Before closure evidence is frozen, sweep the actual public/durable change surface again for suspicious representations, ambiguous naming, unowned contracts, accidental semantic collapse, and hidden assumptions introduced during coding.
$verify-ticket-closure
During the independent adversarial sweep, explicitly challenge whether the candidate introduces an unowned material design/public contract even when all explicit ticket tests/checks pass.
$verify-spec
Before integrated certification is finalized, look for cross-ticket contradictions, duplicated semantics, interface mismatches, lifecycle/identity inconsistencies, or design gaps visible only in composition.
$review-spec
Before review exit, surface material correctness, architecture, naming, coupling, maintainability, or future-constraint concerns even when formal acceptance is otherwise satisfied.
Output
When no material concern remains:
ATTENTION: CLEAR
Checkpoint: <workflow/stage or direct task>
Material concerns: 0ATTENTION: CLEAR certifies only that bounded sweep. Continue the always-on Attention duty immediately after the checkpoint and surface any later material observation even when it contradicts an earlier CLEAR.
When concerns exist:
ATTENTION: FINDINGS
Checkpoint: <workflow/stage or direct task>
1. <concise concern>
Observation: <what was noticed>
Why it matters: <material consequence>
Authority: <resolved by source | unresolved | conflicting>
Recommendation/routing: <what should happen next>
Blocking basis: <existing workflow rule | none>Keep the result compact and decision-useful. Do not bury the owning workflow in a generic review essay.
Deferred Concerns
If a finding is valid but outside the active scope and does not block the current transition, do not fix it opportunistically. Use the repository's established Deferred Future Work Capture policy when the concern is worth preserving durably.
Attention is the sensing mechanism; existing workflow/design/architecture authority decides what happens next.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

