Actions Extensions

Forgejo does not expose Actions run logs or detailed run metadata via its REST API. The proxy fills this gap by scraping the Forgejo web UI using an admin session and returning the data as JSON or plain text. The enriched jobs payload now includes job_index, attempt, log_url, html_url, and steps[] with human-readable durations.

Endpoints

Latest Run

Method	Endpoint	Description
GET	/api/v1/repos/{owner}/{repo}/actions/runs/latest	Get info about the latest run
GET	/api/v1/repos/{owner}/{repo}/actions/runs/latest/logs	Download logs for the latest run
GET	/api/v1/repos/{owner}/{repo}/actions/runs/latest/jobs	List jobs in the latest run
GET	/api/v1/repos/{owner}/{repo}/actions/runs/latest/summary	Summary of the latest run
POST	/api/v1/repos/{owner}/{repo}/actions/runs/latest/rerun	Trigger a rerun of the latest run

Specific Run

Method	Endpoint	Description
GET	/api/v1/repos/{owner}/{repo}/actions/runs/{run_number}/logs	Download logs for a specific run
GET	/api/v1/repos/{owner}/{repo}/actions/runs/{run_number}/jobs	List jobs in a specific run
GET	/api/v1/repos/{owner}/{repo}/actions/runs/{run_number}/summary	Summary of a specific run
POST	/api/v1/repos/{owner}/{repo}/actions/runs/{run_number}/rerun	Trigger a rerun of a specific run

Path Parameters

Parameter	Description
owner	Repository owner (user or organisation name)
repo	Repository name
run_number	The sequential run counter visible in the Forgejo web UI URL (e.g., /actions/runs/42)

Note

run_number is a per-repository sequential counter, not a global ID. It matches the number shown in the Forgejo web UI URL.

Query Parameters

Parameter	Type	Description
job	string | integer	Job name (string) or 0-based job index (integer) within the run (default: 0)
attempt	integer	1-based attempt number for the job (default: 1)
workflow_id	string	Filter runs by workflow file name (e.g., ci.yml)
status	string	Filter runs by status — e.g., success, failure, running, waiting

Note

job accepts either a job name string (e.g., build) or a 0-based integer index (e.g., 0 for the first job, 1 for the second). This was fixed in a recent release — both forms are now supported.

Filtering Runs

The GET .../actions/runs endpoint supports workflow_id and status query parameters to narrow results:

# List all runs for a specific workflow
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs?workflow_id=ci.yml'

# List only failed runs
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs?status=failure'

# Combine filters
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs?workflow_id=ci.yml&status=failure'

Latest Run for a Workflow

Use workflow_id with the latest endpoint to get the most recent run of a specific workflow:

curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/latest?workflow_id=ci.yml'

Examples

# Download logs for the latest run
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/latest/logs'

# Download logs for run #42 by job name
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/42/logs?job=build'

# Download logs for run #42 by job index (second job)
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/42/logs?job=1'

# List jobs in the latest run filtered by workflow
curl -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/latest/jobs?workflow_id=ci.yml'

# Rerun the latest failed run
curl -X POST -H 'Authorization: token $TOKEN' \
  'https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/latest/rerun'

Jobs response shape

Each item from .../jobs is enriched beyond the native Forgejo REST payload:

Field	Meaning
job_index	0-based position of the job within the run
attempt	retry attempt number for the job (1 = first attempt)
log_url	proxy URL for the raw job logs
html_url	upstream Forgejo job-attempt page used to extract step data
steps[]	ordered UI step summaries with summary, status, and optional duration

Example:

[
  {
    "id": 91,
    "name": "build",
    "status": "failure",
    "job_index": 1,
    "attempt": 2,
    "log_url": "https://forgejo-proxy.hochguertel.work/api/v1/repos/myorg/myrepo/actions/runs/42/logs?job=1&attempt=2",
    "html_url": "https://pastoral-oyster.pikapod.net/myorg/myrepo/actions/runs/42/jobs/91?attempt=2",
    "steps": [
      { "summary": "Set up job", "status": "success", "duration": "2s" },
      { "summary": "Run tests", "status": "failure", "duration": "26s" }
    ]
  }
]

The job query parameter used by the logs endpoint accepts the same job_index value or the job name, so you can jump straight from the jobs payload to the matching log stream. The CLI’s thfg run performance command uses this same enriched payload.

How It Works

The proxy authenticates to the Forgejo web UI using an admin session cookie. When a request comes in:

1. The user token is validated against Forgejo’s standard API to confirm read access to the repo
2. The proxy uses its admin session to fetch the run page or log file from the web UI
3. The scraped data is normalized and returned as JSON (or plain text for logs)
4. For jobs, the proxy reads the job-attempt page payload to attach per-step status and duration data