abi/screenshot-to-code · warning · HTTPException

No runs recorded

Error message

No runs recorded

What it means

Raised by _fetch_run() (404) when the SQLite index database returned by get_agent_runs_db_path() does not exist on disk. It means the backend has never recorded any agent run on this machine — the index DB is created lazily on the first run, not at startup. Note the sibling list endpoint GET /agent-runs returns an empty list in the same situation, so the detail endpoints are stricter than the list endpoint.

Source

Thrown at backend/routes/agent_runs.py:139

    return AgentRunSummary(**data)


def _directory_size_bytes(path: str) -> int:
    total = 0
    for root, _, files in os.walk(path):
        for name in files:
            try:
                total += os.path.getsize(os.path.join(root, name))
            except OSError:
                continue
    return total


def _fetch_run(run_id: str) -> AgentRunSummary:
    if not RUN_ID_PATTERN.match(run_id):
        raise HTTPException(status_code=400, detail="Invalid run id")
    if not os.path.isfile(get_agent_runs_db_path()):
        raise HTTPException(status_code=404, detail="No runs recorded")
    conn = open_index_db()
    try:
        row = conn.execute(
            f"SELECT {', '.join(_RUN_COLUMNS)} FROM runs WHERE run_id = ?",
            (run_id,),
        ).fetchone()
    finally:
        conn.close()
    if row is None:
        raise HTTPException(status_code=404, detail="Run not found")
    return _row_to_summary(row)


@router.get("/agent-runs", response_model=AgentRunListResponse)
async def list_agent_runs(limit: int = 200) -> AgentRunListResponse:
    runs_directory = get_agent_runs_directory()
    if not os.path.isfile(get_agent_runs_db_path()):
        return AgentRunListResponse(

View on GitHub (pinned to d026163f58)

Solutions

1. Run one agent generation first so the indexer creates the DB, then retry.
2. Confirm the runs directory/db path the backend uses (call GET /agent-runs and read runs_directory from the response) and check the DB file exists there.
3. If the DB was deleted but run directories still exist, re-run whatever indexing/recording step your version provides, or restore the DB from backup.
4. If you moved the data directory via env var, restart the backend so paths resolve consistently.

Example fix

# before
resp = client.get(f"/agent-runs/{run_id}/output")  # 404 'No runs recorded'

# after
listing = client.get("/agent-runs").json()
if not listing["runs"]:
    print(f"no runs yet under {listing['runs_directory']}; run a generation first")
else:
    resp = client.get(f"/agent-runs/{run_id}/output")

listing = client.get("/agent-runs").json()
if not listing["runs"]:
    # index DB absent; detail calls will 404 with 'No runs recorded'
    raise SystemExit("no runs recorded yet — run a generation first")

def has_recorded_runs(listing: dict) -> bool:
    """True when the agent-runs index exists and holds at least one run."""
    return bool(listing.get("runs"))

try:
    detail = client.get(f"/agent-runs/{run_id}").raise_for_status().json()
except httpx.HTTPStatusError as e:
    if e.response.status_code == 404 and "No runs recorded" in e.response.text:
        detail = None  # fresh install: nothing recorded yet
    else:
        raise

When it happens

Trigger: Calling GET /agent-runs/{run_id} (or /output, /assets/...) on a fresh installation, or after the runs data directory was pruned/deleted, or when SCREENSHOT_TO_CODE_DATA_DIR (or the equivalent env var controlling the runs directory) points somewhere the DB was never created.

Common situations: Fresh clone/first launch before any generation run; the ~/.screenshot-to-code data dir was wiped; a changed data-dir env var between runs; pruning removed the DB file itself.

Related errors

• Run not found
• Run has no captured output
• Asset not found
• Session not found
• Invalid run id

AI-assisted analysis of abi/screenshot-to-code@d026163f58 (2026-08-14). Data as JSON: /api/errors/1136e92dfe0e3b61. Report an issue: GitHub.