Dev Logs
Where every dev log and dev DB lives (one layout, forge env up dev in this checkout or in ../control-plane), how to prove a log sink is live before trusting a zero-hit grep, and Proxyman request correlation. Load before debugging anything in a running dev stack.
mkdir -p ~/.claude/skills/dev-logs && curl -fsSL 'https://raw.githubusercontent.com/reliant-labs/reliant/main/.reliant/skills/dev-logs/SKILL.md' -o ~/.claude/skills/dev-logs/SKILL.md
Get the complete skill folder, including its scripts and reference files.
When to use this skill
Use Dev Logs when an AI agent needs a reusable SKILL.md workflow for this job: Where every dev log and dev DB lives (one layout, forge env up dev in this checkout or in ../control-plane), how to prove a log sink is live before trusting a zero-hit grep, and Proxyman request correlation. Load before debugging anything in a running dev stack.
When not to use it
Skip Dev Logs when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.
How to install
Where will you use it?
mkdir -p ~/.claude/skills/dev-logs && curl -fsSL 'https://raw.githubusercontent.com/reliant-labs/reliant/main/.reliant/skills/dev-logs/SKILL.md' -o ~/.claude/skills/dev-logs/SKILL.md
- Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/dev-logs/SKILL.md. For a shared project, commit the folder at .claude/skills/dev-logs/ instead.
- Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
- For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.
Full install guide for Claude, Cursor, and Codex →
Try a starter prompt
After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.
Use the Dev Logs skill for this task: Use Dev Logs when this documented workflow matches the task. Confirm any required inputs or files before making changes.
What supports this recommendation?
69/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.
- Documentation
- 77/100
- Practical detail
- 68/100
- Source evidence
- 58/100
- Source signals
- 70/100
Model compatibility evidence
- claudeinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- chatgptinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- geminiinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- copilotinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- llamainferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- perplexityinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- mistralinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
- grokinferred
The skill uses model-agnostic prompt or LLM terminology. Portability is inferred and has not been verified on this model.
Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.
What this skill does
# Dev logs and databases
## One layout: `.forge/logs/dev/` of the checkout that ran `forge env up dev`
Every dev stack is a `forge env up dev`, and forge tees every process's stdout — server roles, Vite, the browser console, Electron — into ONE file per workload under the checkout that started it:
``` <checkout>/.forge/logs/dev/<workload>.log ```
There are two checkouts that can run one. Find out which stack produced the behaviour you are debugging before you read anything:
| Stack | Started with | Logs | Its reliant database | | --- | --- | --- | --- | | **reliant's own** (the product alone: api, worker, gateway, web; no control plane, no k3d) | `forge env up dev` in this checkout (`deploy/kcl/dev/`) | `./.forge/logs/dev/` | forge host Postgres 16, on this worktree's port block; database `reliant` | | **control-plane's** (the full platform: billing, LLM gateway, managed workspaces, Electron) | `forge env up dev` in `../control-plane` | `../control-plane/.forge/logs/dev/` | control-plane's docker-compose Postgres on **localhost:5434**, the database its `forge env config dev` names (one per worktree). **Real data: read-only, never drop or mutate it** |
Neither port nor database name is fixed: each worktree's stack gets its own 100-port block (`forge env devstack list`). Ask forge, never a file:
```bash ./.reliant/tools/port-info.sh # this checkout's stack: ports, URLs, database, log dir ./.reliant/tools/port-info.sh --cp # control-plane's forge env config dev --workload reliant-api-server # every variable a process got forge env config dev --json | jq -r '.workloads[].env.DATABASE_URL // empty' | sort -u ```
| Log file | Content | Stack | | --- | --- | --- | | `reliant-temporal-worker.log` | **Activities: tool execution, call_llm, spawns.** Where tool call ids live. Busiest and usually the one you want. | both | | `reliant-api-server.log` | RPC handlers: interrupt, pause, send, queueing | both | | `daemon-gateway.log` | Daemon connections + tool routing | both | | `reliant-migrate.log` | The `reliant db migrate up` job that gates the server roles | both | | `frontend_web.log` | **The UI.** Every browser `console.*` line, prefixed `[browser:<level>]`, plus uncaught errors and unhandled rejections | reliant's | | `frontend_reliant-web.log` | The same, for control-plane's stack — Electron's renderer AND a plain browser tab | control-plane's | | `reliant-electron-main.log` | Electron MAIN process only (Node side: window lifecycle, daemon spawn, IPC). Not the UI. | control-plane's (`-D electron=1`) | | `reliant-electron.log` | The `npm run dev:electron` process's own stdout, as captured by forge | control-plane's | | `admin-server.log` | platform-api (control-plane's public API; the file keeps its old name until the deploy rename ships): proxy, auth, billing, **coupons**, daemon create, CompleteOnboarding | control-plane's |
Host infrastructure (reliant's stack) logs to its own data dir: `.forge/hostinfra/<name>/<engine>.log`.
```bash grep '\[browser:error\]' .forge/logs/dev/*.log # UI errors, this checkout's stack grep -r "$CHAT_ID" ../control-plane/.forge/logs/dev/ # one id across control-plane's stack ./.reliant/tools/log-search.sh --errors # both of those, per file ./.reliant/tools/log-search.sh --cp "$CHAT_ID" ```
Quick disambiguation — run this before trusting any log grep:
```bash C=<chat-id> for f in .forge/logs/dev/*.log ../control-plane/.forge/logs/dev/*.log; do n=$(grep -c "$C" "$f" 2>/dev/null); [ "${n:-0}" -gt 0 ] && echo "$f: $n" done ```
`reliant-dev workflow analyze` and `scripts/wf-supervise` read control-plane's database because they supervise control-plane-backed runs. A chat that is not in one stack's database is almost certainly in the other's: pointing a tool at the wrong stack reads the wrong database, it is not drift.
Tests never touch either database: they boot forge pgtest's embedded Postgres (or `FORGE_TEST_POSTGRES_URL`) and never read `DATABASE_URL`.
## How the browser half gets there — do not add another log location
`frontends/web/src/lib/browser-log-boot.ts` (imported FIRST in `main.tsx`, before any other import can log) wraps `console.*` and POSTs to `/__forge/log`; the Vite plugin in `frontends/web/vite-plugin-browser-logs.ts` prints each line to its own stdout, which forge already captures. It runs in Electron too — the renderer is the same origin as a browser tab, so treating them differently is what previously made the UI go dark with no clue why.
The Electron main process writes to the same directory via `RELIANT_LOG_DIR`, set in `control-plane/deploy/kcl/dev/main.k` (read at process start, so a change needs a stack restart). Unset, it falls back to `.reliant/logs/` for a bare `npm run dev:electron`.
`.reliant/logs/` and the old log files under `data/` were all retired because a second plausible-looking file is worse than none: greps against it return real output while silently missing most of the stream, which is indistinguishable from "my code never ran". A plugin that writes its own file recreates that failure.
## Before concluding "my logging never ran", prove the sink is live
Zero hits means one of two very different things — the code did not execute, or you are reading a file nothing writes to — and they are indistinguishable from the grep alone. Cheapest discriminator, in order:
```bash # 1. Is the file being written RIGHT NOW? (every log's mtime, and the time now) ./.reliant/tools/log-search.sh --where # --cp for control-plane's stack
# 2. Does the running server actually serve your edit? (Vite, no rebuild needed) WEB=$(forge env config dev --json | jq -r '.workloads[] | select(.name == "web") | .env.VITE_APP_URL') curl -s "$WEB/src/path/to/File.tsx" | grep -c 'your-tag'
# 3. Only then: is the code path reached? grep -c 'your-tag' <logfile> ```
Step 2 is the one that gets skipped. It separates "my change is not loaded" from "my change is loaded but that branch never runs" in a single command. Ports are per worktree: `./.reliant/tools/port-info.sh` prints the web URL.
## Proxyman request correlation
For local debug runs, Proxyman can inject a response header carrying its internal flow ID, `x-proxyman-id` (from Proxyman script `context.flow.id`). Debug-only correlation behavior, not a production contract.
1. Find `proxyman_id` / `x-proxyman-id` in the logs: `./.reliant/tools/log-search.sh "proxyman_id\|x-proxyman-id"` 2. Open that flow with Proxyman MCP: `mcp__proxyman__get_flow_detail(flow_id="...")`, or search: `mcp__proxyman__filter_flows(key="responseHeader", matching="contains", value="x-proxyman-id")`.
Intended uses
- Use Dev Logs when this documented workflow matches the task.
Implementation notes
- Review the source instructions and adapt inputs before running the workflow.
Try it, then tell us how it went
A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.