tools/prompts: fix task_add printf bug; point agents at t/idx for tool discovery

printf format string starting with '-' was treated as a flag; use '--' to
terminate option parsing. Agents reading individual tool scripts to understand
them should use $OLLIE/t/idx instead.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

prompts/tools-file.md

@@ -25,6 +25,7 @@ execute_code: steps=[{tool: "file_grep",  args: ["func.*Handler", "src/"]}]
 
 **Constraints**:
 - Never read the full contents of session files (`$OLLIE/s/*/chat`, `$OLLIE/s/*/log`). The conversation history is already in your context window; use `memory_recall` for facts that predate the current session. Grepping these files for a specific term is fine.
+- Never read individual tool scripts to understand their behavior. To discover or verify tool capabilities, read `$OLLIE/t/idx`.
 - `file_write` and `file_edit` require a prior `file_read` in the same session. `file_read` creates a sentinel (keyed by path hash, verified by content hash) that `file_write`/`file_edit` check before proceeding. If the file has been modified outside the session since it was read, the sentinel will not match and you must call `file_read` again before writing.
 - Shell is fine for *processing* the output of file tools — piping results, extracting fields, chaining with other commands. The restriction is on using shell as a substitute for the file tools themselves.
 - Always use absolute paths.

tools/task_add

@@ -13,7 +13,7 @@ current="$(cat "$plan" 2>/dev/null)"
 {
     [ -n "$current" ] && printf '%s\n' "$current"
     for task in "$@"; do
-        printf '- [ ] %s\n' "$task"
+        printf -- '- [ ] %s\n' "$task"
     done
 } > "$plan"