Pixel-CLI/pixel / .agents/skills/pixel-retro
pixel-retro
VerifiedMine the last 24 h (or a given window) of pixel usage — every repo's `.pixel/actions.jsonl` plus the agent transcripts indexed by `pixel recall` (claude, codex, pi…) — for frictions pixel actually caused, verify each one against the current binary and code, and propose ranked, evidence-backed improvements (bug fix, error message, flag, agent prompt, rule, doc); for the pixel repo itself, also measure each pull request from first edit to validated CI and where agents sat idle (`lead_time.py`). Suggests only; implements nothing without the user's pick. Use when the user says "/pixel-retro", "retro pixel", "qu'est-ce qui a coincé avec pixel", "améliorations pixel depuis les transcripts", "pixel friction", "où l'agent attend", "bottleneck", "lead time", or asks what to improve in pixel or its development workflow from recent sessions.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 21.9 KB |
| adherence.py | script | 18.3 KB |
| lead_time.py | script | 22.7 KB |
| ledger.py | script | 1.5 KB |
| roots.py | script | 3.2 KB |
| slow.py | script | 5.6 KB |
| test_adherence.py | script | 9.6 KB |
| test_lead_time.py | script | 17.4 KB |
| test_slow.py | script | 2.4 KB |
roots.py
View on GitHub# SPDX-FileCopyrightText: The Pixel contributors
# SPDX-License-Identifier: MIT
"""Print the repository roots with a .pixel/actions.jsonl among the sessions of a window.
Usage: python3 roots.py <window> (12h, 7d, 3w, or an ISO date)
`pixel recall sessions` caps --limit at 200, has no offset, filters
`ts_last >= since AND ts_first <= until` and orders by ts_last, so no page
cursor is exact. The window is cut into slices instead: a slice that returns
fewer than 200 sessions is complete, a full one is split in two. A session
spanning several slices is counted once (by id). A full slice whose sessions
all span it (halving cannot shrink it), or a spent query budget, ends the run with exit 1 and an
`incomplete:` line, after printing the roots found so far.
"""
import datetime, json, os, re, subprocess, sys
CAP = 200
UTC = datetime.timezone.utc
def parse_window(arg, now):
m = re.fullmatch(r"(\d+)([hdw])", arg)
if m:
hours = {"h": 1, "d": 24, "w": 168}[m.group(2)] * int(m.group(1))
return now - datetime.timedelta(hours=hours)
start = datetime.datetime.fromisoformat(arg)
return start if start.tzinfo else start.replace(tzinfo=UTC)
def sessions(since, until):
cmd = ["pixel", "recall", "sessions", "--since", since.isoformat(),
"--until", until.isoformat(), "--limit", str(CAP), "--subagents",
"--json", "--metrics", "off"]
out = subprocess.run(cmd, capture_output=True, text=True, check=True).stdout
return json.loads(out)["sessions"]
BUDGET = 400 # recall queries; a real 30-day window needed 5
def collect(since, until, seen, incomplete, budget):
if budget[0] <= 0:
incomplete.append(since)
return
budget[0] -= 1
page = sessions(since, until)
seen.update((s["id"], s["cwd"]) for s in page)
if len(page) < CAP:
return
# When every returned session spans the whole slice, both halves return
# the same 200 again: splitting cannot make progress, so flag and stop.
lo, hi = since.timestamp() * 1000, until.timestamp() * 1000
spans_all = all(s["ts_first"] <= lo and s["ts_last"] >= hi for s in page)
if spans_all or until - since <= datetime.timedelta(seconds=1):
incomplete.append(since)
return
mid = since + (until - since) / 2
collect(since, mid, seen, incomplete, budget)
collect(mid, until, seen, incomplete, budget)
now = datetime.datetime.now(UTC)
seen, incomplete = set(), []
budget = [BUDGET]
collect(parse_window(sys.argv[1], now), now, seen, incomplete, budget)
roots = set()
for d in {cwd for _, cwd in seen}:
r = subprocess.run(["git", "-C", d, "rev-parse", "--show-toplevel"],
capture_output=True, text=True)
top = r.stdout.strip()
if r.returncode == 0 and os.path.isfile(os.path.join(top, ".pixel", "actions.jsonl")):
roots.add(top)
print(f"{len(seen)} sessions, {len({c for _, c in seen})} cwds", file=sys.stderr)
print(f"{BUDGET - budget[0]} recall queries", file=sys.stderr)
print("\n".join(sorted(roots)))
if incomplete:
first = min(incomplete).isoformat()
sys.exit(f"incomplete: {len(incomplete)} slice(s) still hold {CAP} sessions "
f"(first at {first}); roots above may miss some repositories")
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |