Issue Intake
SkillDev toolsTurn a noisy inbound request into a bounded intake artifact and an explicit next runx lane.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Issue Intake skill
What this skill tells your AI
The instructions your AI receives, as published by runxhq/runx in skills/issue-intake/SKILL.md and read by ahel’s review.
Convert an inbound thread, support report, or operator request into one explicit intake decision plus the parent change artifact that downstream planning or mutation lanes must share.
This skill does not mutate code, open tickets, or publish replies directly. Its
job is to classify the report, summarize it, draft the next helpful response,
and recommend the next governed lane. That next lane must be explicit:
issue-to-pr, work-plan, reply-only, or manual-review.
In supervisor-style flows, issue-intake is also the commencement gate. It
decides whether work may start at all, whether the next step should stop at a
review comment first, and whether mutation is justified yet. A recommended lane
is not the same thing as build permission.
Use issue-to-pr only when the requested change is bounded enough for one
governed remediation lane. Use work-plan for larger or multi-step
work. Use reply-only when the right answer is guidance rather than mutation.
Use manual-review when the report is ambiguous, risky, or missing key context.
Ground category, severity, and routing in the visible request and supplied
product constraints. Put uncertainty in operator_notes instead of inventing
confidence. The suggested reply should sound like the project owner and lead
with the decision or next action, not read like a ticket macro.
Output Contract
intake_report must contain:
category: one ofbug,feature_request,docs,billing,account,question, orotherseverity: one oflow,medium,high, orcriticalsummary: concise summary of the actual request or reportsuggested_reply: a user-facing reply draft or operator handoff noterecommended_lane:issue-to-pr,work-plan,reply-only, ormanual-reviewrationale: why that lane is the right next stepneeds_human: booleanoperator_notes: array of caveats, missing context, or escalation notes
intake_report may also include supervisor-facing control fields:
commence_decision:approve,hold,reject, orneeds_humanaction_decision:proceed_to_build,proceed_to_plan,request_review, orstopreview_target:thread,outbox_entry, ornonereview_comment: markdown comment body for the supervisor to post before the next lane proceeds
When present, these fields mean:
commence_decisiongates whether the supervisor may start any downstream work at allaction_decision=proceed_to_planmeans the supervisor may open a planning lane such aswork-plan, but still may not start repo mutationaction_decision=request_reviewmeans the supervisor should postreview_commentto the chosenreview_targetand stop there until a later approval or rerun authorizes mutationreview_target=outbox_entryonly makes sense when a current outbox entry already exists. If no draft change, message surface, or other outbox entry exists yet, the supervisor should fall back to the source thread and say that clearly in the posted commentaction_decision=proceed_to_planshould usually still result in a public supervisor comment so the hold/plan decision is visible outside the raw receipt streamrecommended_lane=issue-to-pralone does not authorize a build lane
Always emit change_set alongside intake_report.
Also emit signal when a source event is admitted. signal must follow
runx.signal.v1 and carry the source reference, authenticity or trust level,
dedupe fingerprint, evidence references, and source-thread preview. This packet
is the portable world-before-action state that work-plan, issue-to-pr,
hosted queues, and source-thread projections preserve.
Close the Runx turn after intake is complete with terminal closure control
metadata: a disposition, stable reason code, and concise summary. Closure is
receipt control state, not part of the issue-intake artifact, and must not
pretend the recommended downstream lane has already executed.
When an adapter has provider context beyond the visible thread text, attach it
to signal.evidence_refs or a referenced artifact. Source adapters own
provider-specific fetching and redaction before calling this skill; this skill
only reasons over the supplied, reviewer-safe signal and artifacts.
Hydration is a gate, not a best-effort decoration. If supplied signal or
artifact metadata says provider context is still needed, do not select
action_decision=proceed_to_build. Use manual-review or request_review and
explain the missing adapter context in operator_notes. If provider context is
unavailable, use the remaining signal only when it is still concrete enough for
a bounded reply, plan, or PR; otherwise stop for human review.
The change_set is the parent artifact for any later planning or worker
fanout. It is what keeps multiple repo-scoped lanes aligned to one shared
objective.
change_set must contain:
change_set_idthread_locatorsummarycategoryseverityrecommended_lanecommence_decisionaction_decisiontarget_surfaces: array of objects with:surface: repo, product surface, or bounded target namekind: one ofrepo,package,docs,support, orothermutating: booleanrationale: why this surface is implicated
shared_invariants: array of constraints that all downstream lanes must preservesuccess_criteria: array of concrete outcomes that define success for the whole changeoutbox_entry(optional): current outbox entry for status updates, replies, or draft-change refreshes when the caller already knows it
When recommended_lane=issue-to-pr, also include thread_change_request with:
task_idthread_titlethread_bodythread_locatorthread(optional)outbox_entry(optional)size: one ofmicro,small,medium, orlargerisk: one oflow,medium, orhigh
When recommended_lane=work-plan, also include
workspace_change_plan_request with:
change_set_idobjectiveproject_contextthread_locatorthread(optional)target_surfacesshared_invariantssuccess_criteria
Do not emit both thread_change_request and workspace_change_plan_request for
the same report.
Prefer conservative routing:
- if the report is bounded and well-understood, use
commence_decision=approveandaction_decision=proceed_to_build - if the next step should be planning instead of mutation, use
commence_decision=approveandaction_decision=proceed_to_plan - if the likely next lane is clear but mutation or planning should wait for
maintainer confirmation, use
commence_decision=approveandaction_decision=request_review - if the report is ambiguous, under-specified, or risky, use
commence_decision=holdorneeds_human
Inputs
thread_title: canonical thread titlethread_body: canonical thread body or request textthread_locator(optional): canonical locator for the bounded thread, such as an issue, chat thread, ticket, or local agent sessionthread(optional): provider-backed thread for the current threadoutbox_entry(optional): current outbox entry for replies, draft changes, or refreshessignal(optional): provider-neutralrunx.signal.v1observation gathered by the source adapter before decisionartifact(optional): bounded, reviewer-safe source evidence supplied by the source adapter; provider fetching and redaction must already be completeproduct_context(optional): product-specific constraints or routing hintsoperator_context(optional): maintainer or support posture guidancesource_event(optional): admitted Slack, Sentry, GitHub, file, API, or other provider event. Consuming repos decide source filters before calling this skill.source_policy(optional): source admission and routing policy. Do not hardcode channel names, Sentry projects, or owners in this skill.operational_policy(optional):runx.operational_policy.v1packet used by downstream repo-changing lanes for source, target, runner, and source-thread admission.
Signals
- GitHub stars
- 87
- Forks
- 101
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
issue-intake- Source
- github.com/runxhq/runx