issue-reassess

Name: issue-reassess
Author: apache

68stars52forks

How to install issue-reassess?

apache/magpie/issue-reassess

$npx -y skills add apache/magpie --skill issue-reassess

Installs into the current project.

Summary

Sweep a configured pool of resolved or end-of-life <issue-tracker> issues and re-assess each against the current <default-branch>. Per-issue: invoke issue-reproducer to extract and run the reporter's code, classify the runtime outcome, attach a nature analysis, compose a `v

Files · 1

View on GitHub

SKILL.md

1	<!-- SPDX-License-Identifier: Apache-2.0
2	https://www.apache.org/licenses/LICENSE-2.0 -->
3
4	<!-- Placeholder convention (see ../../AGENTS.md#placeholder-convention-used-in-skill-files):
5	<project-config> → adopter's project-config directory
6	<issue-tracker> → URL of the project's general-issue tracker
7	<issue-tracker-project> → project key within the tracker
8	<upstream> → adopter's public source repo
9	<default-branch> → upstream's default branch (master vs main)
10	<runtime> → recipe for invoking the project's runtime
11	Substitute these with concrete values from the adopting
12	project's <project-config>/ before running any command below. -->
13
14	# issue-reassess
15
16	Use this skill when the task is a campaign over a bounded set
17	of resolved or end-of-life `<issue-tracker>` issues: pick the
18	candidate set, run each reporter's reproducer against
19	`<default-branch>` via [`issue-reproducer`](../issue-reproducer/SKILL.md),
20	classify the outcome, attach a nature analysis, and produce a report
21	a maintainer can scan and act on. The campaign is read-only against
22	the tracker; the output is advisory.
23
24	This skill is the campaign layer. Per-issue mechanics live in
25	sibling skills:
26
27	- [`issue-reproducer`](../issue-reproducer/SKILL.md) — the
28	load-bearing per-issue piece: locate the reproducer, classify the
29	shape, adapt, run, record evidence as `verdict.json`. This skill
30	calls into it for every candidate.
31	- [`issue-triage`](../issue-triage/SKILL.md) — sibling for the
32	unsorted-new pool (this skill handles resolved / EOL / reopened
33	pools instead).
34	- [`issue-fix-workflow`](../issue-fix-workflow/SKILL.md) — where
35	the `still-fails-*` tail goes once the campaign is done; the
36	campaign produces ready-made reproducers for the fix flow.
37	- [`issue-reassess-stats`](../issue-reassess-stats/SKILL.md) —
38	read-only dashboard over the campaign artefacts this skill
39	produces.
40
41	---
42
43	## Golden rules
44
45	Golden rule 1 — read-only on tracker state. Even when 30 of 30
46	findings say `fixed-on-master` with strong evidence, the campaign
47	does not post 30 comments, transition 30 issues, or close
48	anything. The output is a report; a maintainer decides whether and
49	how to publish it. See Transitioning workflow state in
50	[`issue-triage`](../issue-triage/SKILL.md).
51
52	Golden rule 2 — bounded sweeps only. A campaign trying to sweep
53	200 issues in one session blows context, produces low-quality bulk
54	output, and means a crash at issue 150 loses 150 issues' work.
55	Bound the candidate set before the loop starts: a query with a
56	limit, an age bucket, a component slice. Practical first sweeps are
57	5–10 issues; sustained campaigns rarely exceed 50.
58
59	Golden rule 3 — resumable from disk. A 50-issue run that
60	crashes at issue 30 must be resumable from issue 31. Per-issue
61	evidence packages on disk (per
62	[`<project-config>/reproducer-conventions.md`](../../projects/_template/reproducer-conventions.md))
63	are the resumption point — in-memory campaign state is not.
64
65	Golden rule 4 — surface headlines, not stats. *"30
66	fixed-on-master, 5 still-fail, 15 cannot-run"* — the 5 still-fail
67	are usually the most important rows. They're issues where work
68	might actually be done. Surface them at the top of the report; do
69	not bury them under the `fixed-on-master` majority.
70
71	Golden rule 5 — recommend, never decide. *"Close
72	`<KEY>-1234`"* frames the agent as the decider. Phrase as
73	recommendation: *"`fixed-on-master`; a maintainer may want to
74	consider closing after a second pair of eyes."* Workflow decisions
75	belong to maintainers, applied via a separate skill invocation.
76
77	*Golden rule 6 — no fabricated evidence for `cannot-run-`.**
78	"Probably passes on `<default-branch>`." That's a guess in a
79	verdict slot. If it can't be run, the verdict is the `cannot-run-*`
80	category — no further claim. The classification taxonomy has cells
81	for these for a reason

People who liked this also viewed

6 picks

Type

issue-reassess

68stars52forks

View on GitHub

How to install issue-reassess?

apache/magpie/issue-reassess

$npx -y skills add apache/magpie --skill issue-reassess

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	<!-- SPDX-License-Identifier: Apache-2.0
2	https://www.apache.org/licenses/LICENSE-2.0 -->
3
4	<!-- Placeholder convention (see ../../AGENTS.md#placeholder-convention-used-in-skill-files):
5	<project-config> → adopter's project-config directory
6	<issue-tracker> → URL of the project's general-issue tracker
7	<issue-tracker-project> → project key within the tracker
8	<upstream> → adopter's public source repo
9	<default-branch> → upstream's default branch (master vs main)
10	<runtime> → recipe for invoking the project's runtime
11	Substitute these with concrete values from the adopting
12	project's <project-config>/ before running any command below. -->
13
14	# issue-reassess
15
16	Use this skill when the task is a campaign over a bounded set
17	of resolved or end-of-life `<issue-tracker>` issues: pick the
18	candidate set, run each reporter's reproducer against
19	`<default-branch>` via [`issue-reproducer`](../issue-reproducer/SKILL.md),
20	classify the outcome, attach a nature analysis, and produce a report
21	a maintainer can scan and act on. The campaign is read-only against
22	the tracker; the output is advisory.
23
24	This skill is the campaign layer. Per-issue mechanics live in
25	sibling skills:
26
27	- [`issue-reproducer`](../issue-reproducer/SKILL.md) — the
28	load-bearing per-issue piece: locate the reproducer, classify the
29	shape, adapt, run, record evidence as `verdict.json`. This skill
30	calls into it for every candidate.
31	- [`issue-triage`](../issue-triage/SKILL.md) — sibling for the
32	unsorted-new pool (this skill handles resolved / EOL / reopened
33	pools instead).
34	- [`issue-fix-workflow`](../issue-fix-workflow/SKILL.md) — where
35	the `still-fails-*` tail goes once the campaign is done; the
36	campaign produces ready-made reproducers for the fix flow.
37	- [`issue-reassess-stats`](../issue-reassess-stats/SKILL.md) —
38	read-only dashboard over the campaign artefacts this skill
39	produces.
40
41	---
42
43	## Golden rules
44
45	Golden rule 1 — read-only on tracker state. Even when 30 of 30
46	findings say `fixed-on-master` with strong evidence, the campaign
47	does not post 30 comments, transition 30 issues, or close
48	anything. The output is a report; a maintainer decides whether and
49	how to publish it. See Transitioning workflow state in
50	[`issue-triage`](../issue-triage/SKILL.md).
51
52	Golden rule 2 — bounded sweeps only. A campaign trying to sweep
53	200 issues in one session blows context, produces low-quality bulk
54	output, and means a crash at issue 150 loses 150 issues' work.
55	Bound the candidate set before the loop starts: a query with a
56	limit, an age bucket, a component slice. Practical first sweeps are
57	5–10 issues; sustained campaigns rarely exceed 50.
58
59	Golden rule 3 — resumable from disk. A 50-issue run that
60	crashes at issue 30 must be resumable from issue 31. Per-issue
61	evidence packages on disk (per
62	[`<project-config>/reproducer-conventions.md`](../../projects/_template/reproducer-conventions.md))
63	are the resumption point — in-memory campaign state is not.
64
65	Golden rule 4 — surface headlines, not stats. *"30
66	fixed-on-master, 5 still-fail, 15 cannot-run"* — the 5 still-fail
67	are usually the most important rows. They're issues where work
68	might actually be done. Surface them at the top of the report; do
69	not bury them under the `fixed-on-master` majority.
70
71	Golden rule 5 — recommend, never decide. *"Close
72	`<KEY>-1234`"* frames the agent as the decider. Phrase as
73	recommendation: *"`fixed-on-master`; a maintainer may want to
74	consider closing after a second pair of eyes."* Workflow decisions
75	belong to maintainers, applied via a separate skill invocation.
76
77	*Golden rule 6 — no fabricated evidence for `cannot-run-`.**
78	"Probably passes on `<default-branch>`." That's a guess in a
79	verdict slot. If it can't be run, the verdict is the `cannot-run-*`
80	category — no further claim. The classification taxonomy has cells
81	for these for a reason

People who liked this also viewed

6 picks

Type