AGENTS.md: record plain-ASCII default for agent-authored text
New user-preference bullet: agent-authored text avoids typographic
Unicode (em/en dashes, curly quotes, arrows, ellipses) in favour of ASCII
equivalents, with hard enforcement for outward-bound artifacts; documented
exceptions for non-ASCII subject matter, verbatim quotes, and explicit
requests. Existing files are not bulk-retrofitted.
Assisted-by: harness=claude; primary-model=claude-fable-5; api-location=anthropic.com
@ -94,6 +94,7 @@ When the user requests a durable behavior change, record it here or in the relev
- LF line endings are strongly preferred for all files in this repository. Converting a CRLF text file to LF when an edit touches it is correct and welcome - do not preserve CRLF for diff-minimisation. Preserve existing line endings only for files with deliberately mixed/CRLF endings (e.g. line-ending round-trip test data) or when explicitly instructed for a file.
- If the active editor is on a source-derived snapshot, bootstrap copy, or build output path such as `src/bootsupport/`, root `modules/`, root `lib/`, `modules_tcl8/`, `modules_tcl9/`, `lib_tcl8/`, or `lib_tcl9/`, confirm the intended target before editing unless the user explicitly named that path.
- cmd.exe PATH truncation (this machine, and any Windows machine with a heavily populated PATH): cmd.exe truncates a long PATH, so a tool that resolves fine in PowerShell may be "not found" when invoked via `cmd.exe /c`. Use absolute executable paths inside any `cmd /c` command line, and prefer PowerShell-native invocation unless a console host is specifically required (e.g. hidden-console test harnesses). If a tool is missing only under cmd.exe, suspect truncation before absence.
- Agent-authored text is plain ASCII by default: no em/en dashes, curly quotes, arrow or ellipsis characters, or other typographic Unicode - use ASCII equivalents (" - ", straight quotes, "->", "..."). This applies with extra force to outward-bound artifacts (ticket drafts, bug reports, emails, commit messages, anything likely to be pasted into an external system): those must be pure ASCII, verified before handover (e.g. grep for `[^\x00-\x7F]`). Legitimate exceptions: content whose subject matter is itself non-ASCII (encoding/Unicode/ANSI-art test data, or documentation demonstrating such behaviour), verbatim quotes of existing material, and cases where the user explicitly requests non-ASCII. Existing files are not to be bulk-retrofitted - the rule governs newly written text.
- Throwaway fossil repositories (test/experiment repos an agent creates, e.g. in a session scratchpad) must not register in the user's real global fossil config-db: `fossil init`/`fossil open` write persistent `repo:`/`ckout:` rows into `%LOCALAPPDATA%\_fossil`, which is the enumeration source for `dev projects.work` project discovery (G-016/G-017). Set `FOSSIL_HOME` to a disposable scratch directory for the duration of such fossil commands (both fossil and `punk::repo::fossil_get_configdb` honour it first). If pollution has already occurred: `fossil all ignore <repo-path>`, delete the directory, then any `fossil all` command prunes the orphaned `ckout:` row.
- Do not commit new executable binaries (shared libs, .exe, native .so/.dll/.dylib, bare ELF/Mach-O, or zip-based .tm modules embedding executables) to the repository. Existing binaries in `bin/`, `src/vfs/`, `src/vendorlib/`, `src/vendormodules/`, and `src/bootsupport/` are there intentionally pending the build/retrieval infrastructure tracked by goals G-004/G-005/G-006; do not flag, "fix", or hassle the developer about these — they are known and will be removed once G-005 (zig build) or G-006 (pre-built download) provides an alternative. This rule stops agents from adding new binaries; it does not block the developer's interim commits of existing vendor/vfs binaries. It is workflow policy only - deliberately NOT enforced at the local VCS layer: fossil `binary-glob` is `*` (versioned in `.fossil-settings/`) so binary checkins proceed without warnings/prompts, here and in sub-projects like tomlish.