| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
| |
The auth-source-search + funcall-the-secret block was copied four times: calendar-sync--calendar-url, cj/auth-source-secret (ai-config), cj/--auth-source-password (transcription), and cj/slack--get-credential. Each searched authinfo, pulled :secret, and called it when the netrc backend returned a function.
I pulled that into cj/auth-source-secret-value in system-lib (a leaf, so calendar-sync doesn't have to depend on ai-config and drag in the gptel stack). It takes an optional user and returns the secret or nil. The four callers now delegate to it: ai-config layers its required-secret error on top, and the others keep their nil-on-miss behavior. With the direct auth-source-search calls gone, I dropped the now-unused (require 'auth-source) from transcription, slack, and calendar-sync. The helper's autoload covers it.
The transcription tests that exercise the delegated path stay green, and the primitive and the error wrapper get their own tests.
|
| |
|
|
|
|
| |
cj/toggle-gptel and gptel chat errored with "Symbol's function definition is void: gptel-make-anthropic". The local gptel fork on :load-path with :ensure nil ships no generated autoloads, so (require 'gptel) loads gptel.el but never gptel-anthropic.el or gptel-openai.el, where the gptel-make-* constructors live. cj/ensure-gptel-backends then reached gptel-make-anthropic before it was defined.
cj/ensure-gptel-backends now requires gptel-anthropic and gptel-openai first, through a small cj/--gptel-load-backend-libs helper. Verified end-to-end: with the fork on load-path, the constructors are fbound and both backends build.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
I switched the gptel use-package form to `:load-path "~/code/gptel"`
with `:ensure nil` so Emacs loads from the fork instead of the MELPA
release. The fork now carries the narrow `tab-width' copy in
`gptel-org--create-prompt' that karthink redirected the upstream PR
to, which replaces the local `:around' advice on
`gptel--with-buffer-copy-internal' I'd been carrying.
I also dropped the stale test file
`tests/test-ai-config-gptel-prompt-tab-width.el' and the matching
stub in `tests/testutil-ai-config.el'. Both existed only to test the
advice I removed.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
The default-backend swap to gpt-5.5 (commit 0f029ab5) set
`gptel-model' as the string "gpt-5.5". gptel's modeline-display
code calls `symbolp' on the model value and signals
`wrong-type-argument symbolp "gpt-5.5"' on every render, which
manifested as Emacs freezing in the AI-Assistant buffer ("Querying
ChatGPT..." → error in process sentinel → repeated redisplay errors).
Both default-setting sites now use `'gpt-5.5' (interned symbol).
The Anthropic backend tolerated string model names so the original
"claude-opus-4-7" string worked, which is why this hadn't surfaced
before.
|
| |
|
|
|
|
|
|
| |
Two places set the default backend + model on gptel initialization
-- `cj/ensure-gptel-backends' (the lazy-init fallback) and the
`use-package gptel :config' block (the eager-set after initialization).
Both now pick the ChatGPT backend with `gpt-5.5' instead of Claude
with `claude-opus-4-7'.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fourth ADOPT entry from `docs/design/gptel-tools-shortlist.org'.
Lets gptel pull a URL into the conversation so the model can read
docs / current API shapes / etc. without me copy-pasting.
Shape:
- URL must be `http://' or `https://' (file://, ftp://, javascript:,
scheme-less, etc. are rejected at the validator).
- HTML responses go through `pandoc -f html -t plain' so the model
gets a reading shape that isn't full of markup; falls back to
`w3m -dump -T text/html' if pandoc isn't on PATH; signals
`user-error' if neither is. Pass `raw=t' to skip stripping.
- Output capped at 200KB by default, hard cap 1MB; `max_bytes'
argument lets the caller pick a lower cap. Truncation reported
inline.
- 4xx / 5xx response codes signal `error' with the code -- the
alternative is returning an error page body, which the model
would treat as content.
`:confirm t' on the tool because every call is a real outbound
network request. The tool's description warns that URLs go
wherever the user-agent points, including internal networks if
that's what the URL names.
`tests/test-gptel-tools-web-fetch.el' -- 20 tests across Normal /
Boundary / Error. URL validator covers http / https / non-string
/ empty / non-http schemes. `--effective-max-bytes' covers default
/ low-clamp / hard-cap / passthrough. Truncate helper covers
under-cap / at-cap / over-cap with the marker. HTML stripper runs
against real pandoc / w3m (both installed in dev env, neither
should mangle simple markup). Orchestrator stubs
`cj/gptel-web-fetch--retrieve' via `cl-letf' to cover normal /
raw / 4xx / 5xx / oversize / bad-scheme paths.
Wired into `cj/gptel-local-tool-features' so gptel exposes the
tool on next restart.
Note: `call-process-region' invocation flattened to a single
`with-temp-buffer' with DELETE=t -- the initial draft nested a
second temp buffer and routed output to the inner one, which got
killed before `buffer-string' on the outer ran. Test caught it.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Three read-only git context tools so gptel can see what's changed
without me pasting `git status` / `git log` / `git diff` output into
every chat turn. Builds the first batch from the ADOPT bucket in
`docs/design/gptel-tools-shortlist.org`.
Shape per tool:
- `gptel-tools/git_status.el` — `git status --short --branch` for a
directory inside a git working tree under HOME. Returns the
porcelain output, or a "Clean working tree" marker when only the
branch line is present.
- `gptel-tools/git_log.el` — `git log --oneline -nN` with an optional
`--since` filter. N defaults to 20, capped at 100; nil / non-
integer / out-of-range N falls back to the default.
- `gptel-tools/git_diff.el` — `git diff [REF1 [REF2]] [-- FILE]`.
Output capped at ~500KB so a runaway diff can't blow up context;
truncation is reported inline.
Validation is uniform: path must resolve under HOME, must be a
directory, must be inside a git working tree (verified via
`git rev-parse --is-inside-work-tree`). Color is disabled via
`-c color.ui=false` at the git level (`git status` doesn't accept
`--no-color` directly).
Tests run against real temp git repos created via `process-file`,
not mocked — there's nothing in gptel-tools/git_*.el that's
process-mockable in a meaningful way, and a real `git init` + a
couple of commits is cheaper than building a fake. 31 tests total:
7 for git_status, 11 for git_log, 13 for git_diff.
Wired into `cj/gptel-local-tool-features` so gptel exposes the
three tools on next restart.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
conversations
`cj/gptel-load-conversation` prompts via `completing-read`. A
dedicated browser shows what each conversation is about at a
glance and supports single-key load / delete / rename without
having to scroll a minibuffer list.
New module `modules/ai-conversations-browser.el` +
`cj/gptel-browse-conversations` entry point bound to `C-; a b`
("browse conversations"). Opens `*GPTel-Conversations*` in
`cj/gptel-browser-mode` (a `special-mode` derivative).
Each row shows date, time, topic slug, and a preview of the most
recent message (length configurable via
`cj/gptel-browser-preview-length`, default 60 chars). Rows sort
newest first.
In the browser:
- `RET` / `l`: load the conversation (delegates to
`cj/gptel-load-conversation` with the file pre-selected via a
`cl-letf` stub on `completing-read` so the user isn't prompted
twice), then bury the window.
- `d`: delete the file under point after `y-or-n-p` confirmation,
re-render.
- `r`: rename the file under point. Preserves the timestamp,
slugifies the new topic, refuses unchanged input and existing
targets.
- `g`: refresh.
- `n` / `p`: next / previous row.
- `q`: quit-window.
21 tests cover the helpers (topic parsing, header stripping,
preview shaping for truncate / short / empty cases, row-for-file
with conversation + non-conversation filenames, rows enumeration,
render output for empty + populated cases, newest-first sort,
rename-target preservation of timestamp + slug, rename-target
error on missing timestamp) and the file-touching actions (delete
with y, cancel with n, rename, rename-on-empty-line error).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
`gptel-rewrite` is the killer feature for the keep-gptel decision,
and it now lives behind two commands instead of the bare call:
- `cj/gptel-rewrite-with-directive` (`C-; a r`, replacing the
former bare `gptel-rewrite` binding): completing-read on a
directive name from `cj/gptel-rewrite-directives`, then rewrite
the active region.
- `cj/gptel-rewrite-redo-with-different-directive` (`C-; a R`):
replay the prior region with a different directive. The region
is preserved via markers stored buffer-local on the first call so
it survives accept/reject of the prior rewrite.
I picked the hook injection approach over an `:after`-advice +
state-capture pattern. `gptel-rewrite-directives-hook` is an
abnormal hook gptel-rewrite already consults for a per-call
system message. Wrapping the call in a one-shot `let`-binding on
that hook gives the directive exactly the lifetime of the rewrite
and leaves nothing to clean up. Mutating `gptel-directives`
globally would mean either restoring it afterward or living with
the change -- both worse than the hook.
Directives ship inline as a `defcustom` alist with the six names
called out in the task -- `terse`, `fix-grammar`,
`refactor-readability`, `add-docstring`, `explain-as-comment`,
`shorten`. Customization is a `customize-variable` or `setq`
away.
9 tests cover the defcustom shape (default names present, bodies
non-empty strings), the wrapper (normal path, no-region error,
unknown-directive error, last-state recording), and the redo
(replays the prior region, errors when no previous, excludes the
current directive from the re-pick prompt). `gptel-rewrite`
stubbed in tests so no rewrite UI fires.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New module `modules/ai-quick-ask.el`. Bound to `C-; a q` via
`cj/ai-keymap` ("quick ask").
`cj/gptel-quick-ask` reads a prompt in the minibuffer, creates a
transient `*GPTel-Quick*` buffer in `cj/gptel-quick-mode` (a
special-mode derivative with `q` / `escape` / `c` bindings), inserts
"Q: <prompt>" plus a response marker, then calls `gptel-request`
with `:stream t` so the answer streams into the buffer. Doesn't
touch `*AI-Assistant*`, doesn't autosave.
Two follow-up commands work in the buffer:
- `cj/gptel-quick-dismiss` (`q` / `escape`): delete the window and
kill the buffer. Idempotent when the buffer is absent.
- `cj/gptel-quick-continue` (`c`): extract the prompt + response,
seed them into `*AI-Assistant*` under proper org headings (matching
the `cj/gptel--fresh-org-prefix` shape), display the side window,
then dismiss the quick buffer.
13 tests cover the pure helpers (initial-text shape, response
extraction across normal / multi-line / no-marker / empty inputs,
seed-text shape), the ask path (buffer created in right mode,
prompt recorded, gptel-request called, empty-prompt error), the
dismiss path (kills buffer / no-op when absent), and the continue
path (seeds `*AI-Assistant*`, dismisses quick buffer, errors
outside a quick buffer). `gptel-request` is stubbed in tests so
nothing hits the network.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
indicator
`cj/gptel-autosave-enabled` flipped to t inside the save/load entry
points with no way back off short of editing the variable or
clearing the buffer, and no visible indicator that it was on.
Two pieces:
- `cj/gptel-autosave-toggle` flips the buffer-local state in the
current GPTel buffer. Bound to `C-; a A` via `cj/ai-keymap`
(which-key: "toggle autosave"). When autosave is OFF and no
filepath is configured yet, the command prompts to save the
conversation first so a save target exists; otherwise it just
flips the bit.
- `cj/gptel-autosave-mode-line-format` surfaces " [AS]" in the
mode-line when autosave is on, blank when off. Installed via a
`gptel-mode-hook` so every GPTel buffer picks it up. The install
helper is idempotent.
6 new tests cover enable/disable paths, the no-filepath prompt path,
the not-a-gptel-buffer error path, the mode-line format evaluation,
and the install idempotence.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The wiring keyed on `with-eval-after-load 'magit` fires while two
of its three references are still undefined. `magit.el` calls
`(provide 'magit)` BEFORE its `cl-eval-when (load eval)` block
requires `magit-commit` and `magit-stash`. At that moment the
`magit-commit` transient prefix doesn't exist, and
`transient-append-suffix` silently no-ops on missing prefixes
(default `transient-error-on-insert-failure` is nil). The "g
Generate commit" and "x Explain" suffixes never landed. Only the
M-g binding worked, because `git-commit` IS required before
provide.
Three per-feature hooks replace the single `'magit` hook: one each
on `git-commit`, `magit-commit`, and `magit-diff`. Each hooks the
exact dependency the wiring needs, side-stepping the load-order
race entirely.
The companion test was rewritten to check `after-load-alist`
registration rather than drive the hooks through `provide`. Emacs
30 batch mode doesn't fire registered `eval-after-load` callbacks
on `provide` alone -- only an actual `load` does. Inspecting the
registration is the stronger guard anyway: the regression is "a
single `'magit` hook," and the right shape of that check is "no
entry under `magit`, entries under `git-commit`, `magit-commit`,
`magit-diff`."
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I rewrote `update_text_file.el` in pure Elisp. The previous version
shelled out to sed for everything, had a stray quote terminator at EOF
(line 149) that broke loading, produced literal backslash-n where
actual newlines were expected, and prompted via `y-or-n-p` redundantly
with gptel's own `:confirm t` flag.
The five operations -- replace, append, prepend, insert-at-line,
delete-lines -- split into pure string transforms that test without
touching the disk. The file-level wrapper validates the path, enforces
a 10MB size limit, takes a timestamped backup, and writes atomically.
No backup is created when the operation is a no-op. Patterns are
literal substrings (not regex) so the model can't trip over
metacharacter quoting.
`tests/test-update-text-file.el` covers Normal / Boundary / Error per
operation plus the file-level wrapper. 48 tests green. Added
`update_text_file` to `cj/gptel-local-tool-features` so gptel exposes
the tool after restart.
|
| |
|
|
|
|
|
|
| |
gptel's `gptel--with-buffer-copy-internal` copies the source buffer's `major-mode` symbol but doesn't run mode hooks. An inherited-org-mode prompt buffer keeps `tab-width` at this config's global default of 4 instead of the 8 that `org-mode-hook` would set. When gptel later parses the prompt buffer with `org-element`, Org's `tab-width=8` guard raises "Tab width in Org files must be 8, not 4."
I was hitting this on every second `gptel-magit-generate-message` from COMMIT_EDITMSG. `vc-config.el` sets `git-commit-major-mode 'org-mode'`, and the diffs contained list-shaped content that `org-element--list-struct` parsed.
The advice forces `tab-width=8` in the prompt buffer when its inherited mode is org-mode. It's a local workaround for an upstream gap. An upstream patch to run `(delay-mode-hooks (funcall major-mode))` in the buffer-copy is the real fix. I'll send it next.
|
| |
|
|
|
|
|
| |
Replace raw autoload calls with a `use-package` declaration so
`use-package-always-ensure` installs gptel-magit on machines that
haven't run `package-install`, fixing the "Cannot open load file"
error on transient setup.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Anthropic: bump Opus 4.6 → 4.7 (current frontier). Sonnet 4.6
and Haiku 4.5 stay -- still current. Default `gptel-model' setq
also bumped to claude-opus-4-7 in both places it was set.
OpenAI: drop the cohort retired from ChatGPT on 2026-02-13
(gpt-4o, gpt-5 original, gpt-4.1, o1). Replace with the current
lineup: gpt-5.5 (flagship), gpt-5.4-mini (fast/cheap), o3
(reasoning). All three are documented at
developers.openai.com/api/docs/models/.
Drive-by: stale docstring example in cj/gptel--current-model-selection
bumped to match.
gptel's bundled :models list only knows up to May-2025 IDs but
gptel-make-anthropic / gptel-make-openai accept any string and pass
it straight to the API, so newer names work without a gptel upgrade.
|
| |
|
|
| |
Move the local GPTel tool wiring out of init.el and into ai-config. The tools directory and feature list are now configurable, missing optional tools are non-fatal, and focused tests cover the loading behavior.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
F9 was a single command that always opened the project picker. Three small frustrations stacked up. With one claude buffer open and not visible, F9 was a redundant prompt to pick a project that already had a session. With claude visible, there was no way to bury it without M-x quit-window. With two projects' buffers alive, swapping between them was a buffer-switch chore.
F9 is now a dispatch:
- Claude visible in this frame: quit the window (toggle off) and capture the geometry first.
- Exactly one claude buffer alive but hidden: re-display it (DWIM single-buffer case).
- Zero or two-plus alive: fall through to the project picker.
C-F9 is the always-pick-project entry point for explicit project switches. M-F9 is a buffer picker over the alive claude buffers. If a claude window is currently shown, the picked buffer replaces it in that window so the split orientation and size carry over. The shown buffer sorts last in the picker with a [shown] marker so RET picks "the other one."
Split geometry persists across toggles. Two module-level vars (cj/--ai-vterm-last-direction, cj/--ai-vterm-last-size) capture at toggle-off and feed a custom display action. After M-S-t flips claude from right to bottom, F9 toggle-off-then-on returns it at the bottom. After a mouse resize, the next toggle restores that fraction. State is per-session. Restarts reset to default right/0.5.
Two display-buffer fixes came out of testing:
- save-window-excursion around (vterm name) keeps the dashboard from being buried on a fresh F9 at startup. vterm calls pop-to-buffer-same-window internally, which would otherwise replace the selected window's buffer before the alist could route the new one.
- The action chain swaps display-buffer-use-some-window for a more specific cj/--ai-vterm-reuse-existing-claude. The generic version stole non-claude windows on C-F9 when the user was focused inside claude (claude on bottom, code on top -> new project landed in the code window). The specific version only reuses windows that already show a claude buffer.
I reclaimed C-F9 from the gptel toggle in ai-config.el. C-; a t still binds gptel.
I added eight new test files (claude-buffers, displayed-claude-window, dispatch, pick-buffer-candidates, window-geometry, capture-state, display-saved, reuse-existing-claude) plus a regression test on cj/--ai-vterm-show-or-create for the dashboard-preservation fix. All 73 ai-vterm tests pass and the full make test suite is green.
|
| |
|
|
|
|
|
|
| |
The new module picks a Claude-template project from a filtered completing-read list. It scans the same roots the `ai` shell launcher uses, then opens or reuses a vterm buffer named `claude [<repo>]` on the right. F9 launches it. The prior `cj/toggle-gptel` binding moves from F9 to C-F9 so both AI tools share the same physical key.
The display rule chains reuse-window -> use-some-window -> in-direction (right). The resulting window isn't dedicated. That matters because side-window dedication was breaking `buffer-move` (C-M-arrows) and `switch-to-buffer` replacement on the claude buffer. I also narrowed `vterm-toggle`'s display rule to skip `claude [` buffers. Otherwise it claimed them first with its bottom-split + dedicated treatment.
I added 23 tests across 5 files: the buffer-name transform, candidate walker, show-or-create dispatch, picker, and display rule. Design lives at docs/design/ai-vterm.org.
|
| |
|
|
|
|
|
| |
date-to-time used (reverse date) which broke when RRULE UNTIL values
were parsed as 5-element lists (year month day hour minute) from UTC
timestamps. This caused recurring events with UTC UNTIL dates to
expand to 0 occurrences, producing stale calendar entries.
|
| |
|
|
|
|
|
| |
- Extract cj/gptel--build-model-list from cj/gptel-change-model
- Extract cj/gptel--current-model-selection from cj/gptel-change-model
- Add test-ai-config-build-model-list.el (9 tests)
- Add test-ai-config-current-model-selection.el (8 tests)
|
| |
|
|
|
|
|
|
|
|
| |
- Replace use-package gptel-magit hook with autoloads via
with-eval-after-load 'magit (loads gptel only on key press)
- Move org header defuns above use-package to fix load order
- Set gptel-include-reasoning to "*AI-Reasoning*" buffer
- Rebind rewrite to C-; a r, clear context to C-; a c
- Add test-ai-config-gptel-magit-lazy-loading.el (8 tests)
- Mark all ai-config cleanup items DONE in todo.org
|
| |
|
|
|
|
|
|
|
| |
- Move cj/gptel--fresh-org-prefix, cj/gptel--refresh-org-prefix,
cj/gptel-backend-and-model, cj/gptel-insert-model-heading out of
use-package :config to top level (fixes byte-compile warnings)
- Set gptel-include-reasoning to "*AI-Reasoning*" buffer
- Rebind rewrite to C-; a r, clear context to C-; a c
- Update todo.org with completed cleanup items
|
| |
|
|
|
|
|
|
| |
- Add testutil-ai-config.el with gptel stubs for batch testing
- Add tests for cj/gptel--model-to-string (9 tests)
- Add tests for cj/gptel--fresh-org-prefix (8 tests)
- Add tests for cj/gptel-backend-and-model (8 tests)
- Remove dead cj/gptel-backends defvar (duplicates cj/gptel--available-backends)
|
| |
|
|
|
|
|
|
| |
- Escape single quote in cj/ensure-gptel-backends docstring
- Wrap cj/gptel--available-backends docstring to 80 chars
- Add missing docstring to cj/gptel--model-to-string
- Rename cj/gptel--model->string to cj/gptel--model-to-string
- Mark stale model list task as DONE in todo.org
|
| |
|
|
|
|
| |
Remove redundant gptel-backend setq and orphaned section header.
Fix commentary keybinding references: M-a → C-; a, correct buffer
add (.) and rewrite (&) bindings.
|
| |
|
|
|
|
|
|
| |
Update Anthropic model list to current: claude-opus-4-6, claude-sonnet-4-6,
claude-haiku-4-5-20251001. Fix gptel--system-message not picking up the
custom default.org directive (defvar set at load time before gptel-prompts
replaces the default entry). Add cleanup tasks for ai-config, calibredb,
and slack reaction workflow to todo.org.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Enhance which-key integration by providing detailed descriptions for
new key bindings across multiple modules. This improves the
usability of custom keymaps by clarifying the purpose of each
keybinding, making it easier for users to navigate and understand
different menus and options available within the configuration.
This update ensures that all custom keymaps now display a
descriptive label in the which-key popup to explain their
functionality, aiding users in identifying keymap purposes promptly.
|
| |
|
|
|
|
|
|
| |
Optimized key binding configurations across modules for consistency
and reduced redundancy. Improved conditional requiring to handle
errors gracefully in `music-config.el`, ensuring robustness across
different machine environments. Eliminated comments clutter and
adjusted function definitions to adhere to revised standards.
|
| |
|
|
|
| |
Switch key bindings from define-key to the keymap-set function for
improved clarity and modern syntax within the AI operations keymap.
|
| |
|
