1c Project Init

/1c-project-init — Initialize or enrich 1C project

Source workspace: C:\Users\YOUR_USERNAME\workspace\ai\1c-AI-workspace

Mode detection

• .claude/skills/ exists in target → enrich (sync missing/outdated skills & docs)
• No .claude/skills/ → new (full init)

CRITICAL: Use skills, not manual 1C commands

NEVER run 1cv8 DESIGNER manually when a skill exists for the operation.

The /db-dump-xml skill reads .v8-project.json automatically and connects to the remote server directly — no need to run DESIGNER inside the container. License PROF on the local machine is sufficient for all DESIGNER operations including DumpConfigToFiles and extension dumps.

Steps

1. Determine target path

• If argument provided → use it
• Otherwise → current working directory (pwd)

2. Run init script

powershell.exe -NoProfile -File "C:\Users\YOUR_USERNAME\workspace\ai\1c-AI-workspace\.claude\skills\1c-project-init\scripts\init.ps1" -TargetPath "<target>" -Mode <new|enrich>

Script output lists what was copied/updated.

3. If mode = new — collect project info interactively

Ask ALL questions upfront in a single AskUserQuestion call (multiple questions at once).

Question order matters — project type and extension names come FIRST:

1. Project type (options): configuration / extension / configuration + extension / external-processor / other
2. Extension name(s) — ask only if type includes extension. If multiple, comma-separated (e.g. МоёРасширение, ДругоеРасширение)
3. 1C base name (e.g. Forte, Buh)
4. Platform version: 8.3.24 / 8.3.25 / 8.3.27
5. Web publication name (e.g. forte, buh) — for HTTP API and playwright
6. Admin login and password — for .v8-project.json

Note: questions 1+2 can be asked together in first call; 3-6 in second call if user answered that extensions are involved. Or ask all in one call if possible.

Platform → server mapping (all on CT107 / YOUR_EDT_SERVER):

• 8.3.24 → server YOUR_EDT_SERVER:1641, container onec-server-24, web port 8081
• 8.3.25 → server YOUR_EDT_SERVER:1541, container onec-server-25, web port 8080
• 8.3.27 → server YOUR_EDT_SERVER:1741, container onec-server-27, web port 8082

4. Generate CLAUDE.md

Read template: C:\Users\YOUR_USERNAME\workspace\ai\1c-AI-workspace\.claude\skills\1c-project-init\templates\CLAUDE.md.template Fill all placeholders:

• {{PROJECT_NAME}} — project name
• {{PROJECT_DESCRIPTION}} — brief description (include project type + extensions if applicable)
• {{V8_VERSION}} — platform version (e.g. 8.3.25)
• {{SERVER}} — YOUR_EDT_SERVER
• {{PORT}} — from platform map
• {{SERVER_SUFFIX}} — 24 / 25 / 27
• {{WEB_PORT}} — from platform map
• {{BASE_NAME}} — 1C base name
• {{PUBLICATION}} — web publication name
• {{EXTENSION_LINE}} — - Extensions: <names> if extensions present, else omit line
• {{DEV_ZONE}} — src/ / ext/<ExtName>/ / src/ + ext/<ExtName>/
• {{DEV_ZONE_NOTE}} — All modifications go into the extension, not the base configuration. if extension-only or mixed; empty if configuration-only
• {{STRUCTURE_LINES}} — include src/ line if has config, ext/ line if has extensions:
  • config: src/ — configuration source
  • extension: ext/<ExtName>/ — extension source (primary development zone)
  • both: src/ — base configuration source (read-only reference)\next/<ExtName>/ — extension source (PRIMARY development zone)

Write to <target>/CLAUDE.md.

5. Generate .mcp.json

Read template: C:\Users\YOUR_USERNAME\workspace\ai\1c-AI-workspace\.claude\skills\1c-project-init\templates\mcp.json.template Fill all placeholders:

• {{PROJECT_NAME}} — project directory name (for bsl-lsp container)
• {{PUBLICATION}} — web publication name (for ai-debug URL)
• {{WEB_PORT}} — from platform map (8081/8080/8082)
• {{ADMIN_USER}} — admin login
• {{ADMIN_PWD}} — admin password
• {{CODEMETADATA_URL}} — codemetadata MCP URL. Format: http://YOUR_MCP_SERVER:<PORT>/mcp. Port from registry: next free in 7530+ range. If codemetadata not yet deployed, use placeholder http://YOUR_MCP_SERVER:{{CODEMETADATA_PORT}}/mcp and note in report.

Write to <target>/.mcp.json.

6. Generate .v8-project.json if not exists

Read template: C:\Users\YOUR_USERNAME\workspace\ai\1c-AI-workspace\.claude\skills\1c-project-init\templates\v8-project.json.template Fill all placeholders:

• {{V8PATH}} — C:\\Program Files\\1cv8\\<V8_VERSION>\\bin matching the server platform. NEVER leave empty — local default may be a different version.
• {{PORT}} — from platform map (1641/1541/1741)
• {{BASE_NAME}} — 1C base name
• {{ADMIN_USER}} — admin login
• {{ADMIN_PWD}} — admin password
• {{WEB_PORT}} — from platform map (8081/8080/8082)
• {{PUBLICATION}} — web publication name

Write to <target>/.v8-project.json.

7. Create openspec structure if mode = new

openspec/
  project.md        ← project context for AI
  changes/          ← active proposals
  specs/            ← feature specs
  archive/          ← done
dist/               ← built artifacts (CFE/CF/EPF) — tracked in git for distribution

8. Deploy infobase on CT107 (mode = new)

Ask: "Развернуть базу на CT107?"

If yes:

Step 8.1 — Copy .dt file to CT107 (/mnt/data/data-24/ maps to /var/lib/onec/ inside container):

scp "<dt_file_path>" root@YOUR_EDT_SERVER:/mnt/data/data-<SERVER_SUFFIX>/<BASE_NAME>.dt

Step 8.2 — Ensure Xvfb is running in the target container (required for 1cv8 DESIGNER):

ssh root@YOUR_EDT_SERVER "docker exec onec-server-<SERVER_SUFFIX> pgrep Xvfb || docker exec -d onec-server-<SERVER_SUFFIX> Xvfb :99 -screen 0 1024x768x16"

deploy-infobase.sh uses DISPLAY=:99 but does NOT auto-start Xvfb. Without this step, deploy fails with "Unable to initialize GTK+".

Step 8.2b — Run deploy script on CT107:

ssh root@YOUR_EDT_SERVER "/opt/1c-dev/deploy-infobase.sh --name <BASE_NAME> --dt /mnt/data/data-<SERVER_SUFFIX>/<BASE_NAME>.dt --server <SERVER_SUFFIX> --admin-user <user> --admin-pwd <pwd>"

Script /opt/1c-dev/deploy-infobase.sh — universal script (supports 24/25/27):

• --server 24|25|27 — selects container (onec-server-XX), ports, web container automatically
• Auto-detects platform version (ls /opt/1cv8/x86_64/ inside container)
• Auto-detects cluster ID via rac cluster list
• DT path mapping: /mnt/data/data-<SERVER>/X.dt → /var/lib/onec/X.dt inside container

Does 4 steps automatically:

1. Check if infobase already exists (rac infobase summary list); if not — create via rac infobase create + PostgreSQL DB (--create-database). Fails with error if UUID can't be obtained.
2. 1cv8 DESIGNER /RestoreIB — restores .dt. Verifies "успешно завершена" in log, exits on failure.
3. onec-webinst.sh publish — publishes web in onec-web-<SERVER_SUFFIX>
4. rac infobase update --license-distribution=allow

Script behaviour:

• DB password read from /opt/1c-dev/secrets/pg_password (not hardcoded)
• Lock file /tmp/deploy-<NAME>.lock prevents concurrent runs
• Log written to /tmp/deploy-<NAME>-<timestamp>.log (timestamped, not overwritten)
• set -euo pipefail — stops on any error

Known cluster IDs: 8.3.24=eecfed03-e569-4bc7-a729-b8518e4b6859, 8.3.27=0c2ac9c1-8017-4ac8-8b93-35f1409a3ecb. Result URL: http://YOUR_EDT_SERVER:<WEB_PORT>/<BASE_NAME>/

Step 8.3 — Verify infobase availability:

# 1. Infobase registered in cluster
ssh root@YOUR_EDT_SERVER "docker exec onec-server-24 /opt/1cv8/x86_64/<ONEC_VERSION>/rac localhost:1645 infobase summary --cluster=<CLUSTER_ID> | grep -i '<BASE_NAME>'"

# 2. Web publication responds
curl -s -o /dev/null -w "%{http_code}" "http://YOUR_EDT_SERVER:<WEB_PORT>/<BASE_NAME>/"
# Expected: 200 or 302

# 3. HTTP API available (if 1c-ai-debug configured)
curl -s "http://YOUR_EDT_SERVER:<WEB_PORT>/<BASE_NAME>/hs/ai/ping" -u "<user>:<pwd>"
# Expected: {"status":"ok"} or any 200 response

Only proceed to next steps if web check returns 200/302. If 503/404 — check deploy log on CT107: docker exec onec-server-24 cat /var/lib/onec/restore_<BASE_NAME>.log