wait-ci

Name: wait-ci
Author: Codagent-AI

28stars1forks

How to install wait-ci?

codagent-ai/agent-skills/wait-ci

$npx -y skills add Codagent-AI/agent-skills --skill wait-ci

Installs into the current project.

Summary

Polls CI check status for the current branch's pull request and reports pass/fail/pending/comments, surfacing PR review comments even when CI is green. Use when the user says "wait for CI", "check CI", "poll CI", or invokes "codagent:wait-ci".

Files · 1

View on GitHub

SKILL.md

1	# codagent:wait-ci
2
3	Poll CI check status for the current branch's PR and report the result, enriching failures with log output and checking for blocking reviews.
4
5	Use the bundled scripts for all API calls — they encode the correct field names, jq patterns, and GraphQL queries. Do not rewrite API calls inline.
6
7	> NEVER pause to ask the user for permission to wait. Always run the full polling duration silently. Asking mid-execution breaks automation and is never needed — the caller already decided to invoke this skill.
8
9	## How polling works
10
11	`check-ci.sh` is a single-invocation script. It polls every 10 seconds for up to 90 seconds, then exits. Claude calls it in a loop to cover the full max wait time. This keeps each Bash call bounded to ~2 minutes rather than blocking the agent for the full duration.
12
13	Each Bash call to `check-ci.sh` must use `timeout: 120000` (2 minutes).
14
15	## Steps
16
17	### 1. Run the polling loop
18
19	Compute `max_runs = ceil(max_minutes * 60 / 90)`, defaulting to `max_minutes = 15` → 10 runs.
20
21	The skill accepts an optional `--max-minutes N` argument; pass it through to adjust `max_runs`.
22
23	Track `ever_had_checks = false`, `run = 0`, and the overall deadline across iterations. The deadline is `max_minutes` after polling starts and also bounds any review-bot waiting in Step 3.
24
25	For each run:
26
27	```bash
28	bash skills/wait-ci/scripts/check-ci.sh
29	```
30
31	Use `timeout: 120000` on the Bash call. Check the exit code:
32
33	\| Exit code \| Meaning \| Action \|
34	\|---\|---\|---\|
35	\| `0` \| Terminal (`passed` or `failed`) \| Break out of loop, proceed to Step 2 \|
36	\| `2` \| Not yet terminal (`pending` or `no_checks`) \| If `run < max_runs`, re-run; else report timeout \|
37	\| `1` \| Fatal error \| Report error and stop \|
38
39	After each exit-2 result, set `ever_had_checks = ever_had_checks OR result.had_checks`.
40
41	Timeout handling: When all runs are exhausted (exit code 2 on the last run):
42	- If `ever_had_checks` is false → treat CI as non-blocking, then still run Step 3 before reporting a final status
43	- Otherwise → report `pending` with the list of still-running checks
44
45	The script outputs a JSON object with these fields:
46
47	\| Field \| Type \| Description \|
48	\|---\|---\|---\|
49	\| `status` \| string \| `passed`, `failed`, `pending`, `no_checks`, or `comments` (set by caller) \|
50	\| `pr_url` \| string \| PR URL \|
51	\| `pr_number` \| number \| PR number \|
52	\| `head_sha` \| string \| PR head commit queried for this invocation; preserve it when upgrading the status to `comments` \|
53	\| `owner` / `repo` \| string \| Repo coordinates for subsequent calls \|
54	\| `had_checks` \| bool \| Whether any checks were seen on this invocation \|
55	\| `failed_checks` \| array \| Checks with `bucket == "fail"` \|
56	\| `passed_checks` \| array \| Checks with `bucket == "pass"` \|
57	\| `pending_checks` \| array \| Checks still running \|
58	\| `blocking_reviews` \| array \| Reviews with `CHANGES_REQUESTED` (latest per reviewer) \|
59	\| `failed_run_ids` \| array \| GitHub Actions run IDs extracted from failed check links \|
60
61	### 2. Fetch failure logs (if `status == "failed"`)
62
63	For each run ID in `failed_run_ids`:
64
65	```bash
66	gh run view <run-id> --log-failed
67	```
68
69	Keep the last 100 lines if output is longer. External checks (no run ID) get no logs.
70
71	### 3. Gather PR comments (when checks are terminal or no checks exist)
72
73	```bash
74	bash skills/wait-ci/scripts/get-pr-comments.sh <owner> <repo> <pr-number> [<pr-author-login>]
75	```
76
77	Uses GraphQL to check `isResolved` on review threads directly — no jq `!=` workarounds needed.
78
79	Output fields:
80
81	\| Field \| Type \| Description \|
82	\|---\|---\|---\|
83	\| `has_comments` \| bool \| True if any unaddressed comments exist \|
84	\| `unresolved_threads` \| array \| `{file, line, author, body}` per unresolved review thread \|
85	\| `issue_comments` \| array \| `{author, body}` blocking top-level human comments (excluding PR creator) \|
86	\| `informational_bot_comments` \| array \| `{author, body}` non-blocking top-level bot comments retained as evidence \|
87
88	Unresolved review threads are blocking regardless of author. Top-level bot comments do not set `has_comments`; only unresolved threads and human top-level comments do.
89
90	After checks are terminal, inspect `informational_bot_comments`. If a bot explicitly reports that its review is pending or in progress, or asks the caller to check back, wait 10 seconds and call `get-pr-comments.sh` again while time remains before the overall deadline. Re-evaluate the latest evidence on every poll. Do not hard-code bot vendors or exact phrases; use the comment's meaning. Completed summaries and ordinary status notices remain informational and do not delay completion.
91
92	If a bot still explicitly reports unfinished review work when the overall deadline is reached, report `pending`. Preserve its comment in the

People who liked this also viewed

6 picks

Type

wait-ci

28stars1forks

View on GitHub

How to install wait-ci?

codagent-ai/agent-skills/wait-ci

$npx -y skills add Codagent-AI/agent-skills --skill wait-ci

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# codagent:wait-ci
2
3	Poll CI check status for the current branch's PR and report the result, enriching failures with log output and checking for blocking reviews.
4
5	Use the bundled scripts for all API calls — they encode the correct field names, jq patterns, and GraphQL queries. Do not rewrite API calls inline.
6
7	> NEVER pause to ask the user for permission to wait. Always run the full polling duration silently. Asking mid-execution breaks automation and is never needed — the caller already decided to invoke this skill.
8
9	## How polling works
10
11	`check-ci.sh` is a single-invocation script. It polls every 10 seconds for up to 90 seconds, then exits. Claude calls it in a loop to cover the full max wait time. This keeps each Bash call bounded to ~2 minutes rather than blocking the agent for the full duration.
12
13	Each Bash call to `check-ci.sh` must use `timeout: 120000` (2 minutes).
14
15	## Steps
16
17	### 1. Run the polling loop
18
19	Compute `max_runs = ceil(max_minutes * 60 / 90)`, defaulting to `max_minutes = 15` → 10 runs.
20
21	The skill accepts an optional `--max-minutes N` argument; pass it through to adjust `max_runs`.
22
23	Track `ever_had_checks = false`, `run = 0`, and the overall deadline across iterations. The deadline is `max_minutes` after polling starts and also bounds any review-bot waiting in Step 3.
24
25	For each run:
26
27	```bash
28	bash skills/wait-ci/scripts/check-ci.sh
29	```
30
31	Use `timeout: 120000` on the Bash call. Check the exit code:
32
33	\| Exit code \| Meaning \| Action \|
34	\|---\|---\|---\|
35	\| `0` \| Terminal (`passed` or `failed`) \| Break out of loop, proceed to Step 2 \|
36	\| `2` \| Not yet terminal (`pending` or `no_checks`) \| If `run < max_runs`, re-run; else report timeout \|
37	\| `1` \| Fatal error \| Report error and stop \|
38
39	After each exit-2 result, set `ever_had_checks = ever_had_checks OR result.had_checks`.
40
41	Timeout handling: When all runs are exhausted (exit code 2 on the last run):
42	- If `ever_had_checks` is false → treat CI as non-blocking, then still run Step 3 before reporting a final status
43	- Otherwise → report `pending` with the list of still-running checks
44
45	The script outputs a JSON object with these fields:
46
47	\| Field \| Type \| Description \|
48	\|---\|---\|---\|
49	\| `status` \| string \| `passed`, `failed`, `pending`, `no_checks`, or `comments` (set by caller) \|
50	\| `pr_url` \| string \| PR URL \|
51	\| `pr_number` \| number \| PR number \|
52	\| `head_sha` \| string \| PR head commit queried for this invocation; preserve it when upgrading the status to `comments` \|
53	\| `owner` / `repo` \| string \| Repo coordinates for subsequent calls \|
54	\| `had_checks` \| bool \| Whether any checks were seen on this invocation \|
55	\| `failed_checks` \| array \| Checks with `bucket == "fail"` \|
56	\| `passed_checks` \| array \| Checks with `bucket == "pass"` \|
57	\| `pending_checks` \| array \| Checks still running \|
58	\| `blocking_reviews` \| array \| Reviews with `CHANGES_REQUESTED` (latest per reviewer) \|
59	\| `failed_run_ids` \| array \| GitHub Actions run IDs extracted from failed check links \|
60
61	### 2. Fetch failure logs (if `status == "failed"`)
62
63	For each run ID in `failed_run_ids`:
64
65	```bash
66	gh run view <run-id> --log-failed
67	```
68
69	Keep the last 100 lines if output is longer. External checks (no run ID) get no logs.
70
71	### 3. Gather PR comments (when checks are terminal or no checks exist)
72
73	```bash
74	bash skills/wait-ci/scripts/get-pr-comments.sh <owner> <repo> <pr-number> [<pr-author-login>]
75	```
76
77	Uses GraphQL to check `isResolved` on review threads directly — no jq `!=` workarounds needed.
78
79	Output fields:
80
81	\| Field \| Type \| Description \|
82	\|---\|---\|---\|
83	\| `has_comments` \| bool \| True if any unaddressed comments exist \|
84	\| `unresolved_threads` \| array \| `{file, line, author, body}` per unresolved review thread \|
85	\| `issue_comments` \| array \| `{author, body}` blocking top-level human comments (excluding PR creator) \|
86	\| `informational_bot_comments` \| array \| `{author, body}` non-blocking top-level bot comments retained as evidence \|
87
88	Unresolved review threads are blocking regardless of author. Top-level bot comments do not set `has_comments`; only unresolved threads and human top-level comments do.
89
90	After checks are terminal, inspect `informational_bot_comments`. If a bot explicitly reports that its review is pending or in progress, or asks the caller to check back, wait 10 seconds and call `get-pr-comments.sh` again while time remains before the overall deadline. Re-evaluate the latest evidence on every poll. Do not hard-code bot vendors or exact phrases; use the comment's meaning. Completed summaries and ordinary status notices remain informational and do not delay completion.
91
92	If a bot still explicitly reports unfinished review work when the overall deadline is reached, report `pending`. Preserve its comment in the

People who liked this also viewed

6 picks

Type