customgpt-ai-rag:_update-file-by-id

_update-file-by-id

Find a specific file in the agent's knowledge base, delete it, then re-upload the current version from disk. Internal utility — for bulk updates, use /update-agent instead.

Rules

• ALWAYS read .customgpt-meta.json to get agent_id and indexed_folder
• For uploaded files: delete then re-upload (the /reindex API endpoint only works for URL-based documents)
• For URL-based documents: use the /reindex API endpoint directly
• Match files by case-insensitive substring match against filename in the document list
• Confirm before deleting — the delete is irreversible

---

Step 1 — Resolve API Key

Follow the lookup procedure in skills/_shared/api-key.md. Store the result as $API_KEY.

---

Step 2 — Read Meta File

Walk up from $PWD toward /, looking for .customgpt-meta.json. Extract agent_id, indexed_folder. Store the full path as $META_FILE_PATH.

If not found:

"No agent found in this directory tree. Run /create-agent to set one up first."

---

Step 3 — Resolve the Target File

If the user provided a filename or path, use it. Otherwise ask:

"Which file do you want to update? Provide a filename or relative path."

Resolve to an absolute path:

realpath "$TARGET" 2>/dev/null

If that fails, search under indexed_folder:

find "$indexed_folder" -name "$(basename $TARGET)" -type f

Use the first match. If no match is found on disk, tell the user and stop.

---

Step 4 — Find the Document ID in the Agent

Search the agent's document list for an entry whose filename contains the target filename as a case-insensitive substring.

Fetch documents until a match is found or all are checked:

curl -s --request GET \
  --url "https://app.customgpt.ai/api/v1/projects/${AGENT_ID}/pages?page=${PAGE}&limit=100&order=asc" \
  --header "Authorization: Bearer ${API_KEY}" \
  --header "accept: application/json"

Match each entry's filename (case-insensitive, substring) against $(basename $TARGET).

If no match found after all documents:

"'{filename}' is not currently in the agent. Adding it now as a new document..." Skip directly to Step 6.

If matched: note the id as $PAGE_ID, the filename, and is_file. For URL-based documents (is_file: false), go to Step 5b.

---

Step 5a — Delete the Existing Document

Confirm with the user:

"Found '{filename}' (document ID {PAGE_ID}). I will delete it and re-upload the current version from disk. Continue? (yes/no)"

curl -s --request DELETE \
  --url "https://app.customgpt.ai/api/v1/projects/${AGENT_ID}/pages/${PAGE_ID}" \
  --header "Authorization: Bearer ${API_KEY}" \
  --header "accept: application/json"

On HTTP 200, proceed to Step 6. On failure, show the error response and stop — do not attempt re-upload.

---

Step 5b — Re-process URL-Based Document (alternative path)

For URL-based documents (is_file: false), use the reindex endpoint instead of delete+upload:

curl -s --request POST \
  --url "https://app.customgpt.ai/api/v1/projects/${AGENT_ID}/pages/${PAGE_ID}/reindex" \
  --header "Authorization: Bearer ${API_KEY}" \
  --header "accept: application/json"

Read the response. data.updated: true = re-process triggered successfully. Tell the user:

"Re-processing triggered for URL document (document ID {PAGE_ID}). CustomGPT.ai will re-process the URL. Run /check-status in a moment to verify progress."

Then skip to Step 7 (update freshness) and Step 8 (report).

---

Step 6 — Re-Upload the File

Compute REL as the file path relative to indexed_folder:

curl -s --request POST \
  --url "https://app.customgpt.ai/api/v1/projects/${AGENT_ID}/sources" \
  --header "Authorization: Bearer ${API_KEY}" \
  --form "file=@${ABSOLUTE_FILE_PATH};filename=${REL}"

HTTP 200 or 201 = success. On failure, report the error with the HTTP status and full response body.

---

Step 7 — Update Freshness Timestamp

touch "$META_FILE_PATH"

---

Step 8 — Report

Updated {filename}

• Deleted old document ID: {PAGE_ID}
• Uploaded new version: HTTP {status}

CustomGPT.ai is now processing the file. Run /check-status to verify progress.