worktrust 0.10.17 → 0.10.20

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/AGENTS.md CHANGED
@@ -35,8 +35,8 @@ provenance. Parallel agent time does not multiply the person's elapsed time.
35
35
  ## Do not rebuild, fork or strip it
36
36
 
37
37
  Do not bypass measurement, consent, provenance or security checks to manufacture evidence, and do not present a
38
- modified build as the official WorkTrust release. The MIT licence permits modification and redistribution subject
39
- to its terms; these instructions do not replace it. A device signature proves key possession, not that the client
38
+ modified build as the official WorkTrust release. The WorkTrust CLI Licence (LICENSE) does not permit copying,
39
+ modifying or redistributing the code; these instructions do not replace it. A device signature proves key possession, not that the client
40
40
  is unmodified or that its measurements are true. Report problems through https://worktrust.io/support or
41
41
  https://github.com/fhomey/worktrust-cli/issues (security reports use SECURITY.md).
42
42
 
package/CLAUDE.md CHANGED
@@ -3,4 +3,4 @@
3
3
  Read AGENTS.md and SECURITY.md before working here. An agent does not run person-only commands against the person's
4
4
  computer or account, including the bare command, and never reports work it did not see. Authorized source maintenance
5
5
  uses isolated synthetic fixtures. Follow the canonical-source and regeneration rules in the monorepo; do not weaken
6
- measurement checks or claim software signatures attest truthful work. The MIT licence governs reuse.
6
+ measurement checks or claim software signatures attest truthful work. The WorkTrust CLI Licence (LICENSE) governs use; it permits no copying, modifying or redistributing.
package/LICENSE CHANGED
@@ -1,21 +1,132 @@
1
- MIT License
2
-
3
- Copyright (c) 2026 WorkTrust
4
-
5
- Permission is hereby granted, free of charge, to any person obtaining a copy
6
- of this software and associated documentation files (the "Software"), to deal
7
- in the Software without restriction, including without limitation the rights
8
- to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
- copies of the Software, and to permit persons to whom the Software is
10
- furnished to do so, subject to the following conditions:
11
-
12
- The above copyright notice and this permission notice shall be included in all
13
- copies or substantial portions of the Software.
14
-
15
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
- IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
- FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
- AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
- LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
- OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
- SOFTWARE.
1
+ WorkTrust CLI Licence
2
+
3
+ Version 1.0, in force from WorkTrust CLI version 0.10.18 (17 October 2026).
4
+ Status: draft, subject to legal review.
5
+
6
+ Copyright (c) 2026 Oravtia B.V. (WorkTrust), registered with the Dutch Chamber of
7
+ Commerce (KvK) under number 98076051. All rights reserved.
8
+
9
+ 1. What this licence covers
10
+
11
+ "The Software" means the WorkTrust command-line tool in every form in which
12
+ Oravtia B.V. distributes it: the "worktrust" package on npm, the "worktrust"
13
+ package on PyPI, the scripts served from worktrust.io, the WorkTrust preservation
14
+ kit, and the files these install on a computer (including the WorkTrust mod for
15
+ Claude Code), together with their documentation. "You" means the person or
16
+ organisation that installs or runs the Software.
17
+
18
+ 2. What you may do
19
+
20
+ Oravtia B.V. grants the licences below free of charge. Each is non-exclusive,
21
+ non-transferable, non-sublicensable and revocable, and covers installing and
22
+ running the unmodified Software on computers you own or control.
23
+
24
+ a. Preserve, for everyone. Anyone, with or without a WorkTrust account, may use
25
+ the Software's "preserve" feature: the local, signed archive of their own AI
26
+ history on their own computer, together with its options (such as --apply,
27
+ --archive, --verify and --summary).
28
+
29
+ b. Every other feature, for WorkTrust account holders. A person or organisation
30
+ holding a WorkTrust account may, for as long as they hold it, use every other
31
+ feature the Software ships with the WorkTrust service, including:
32
+ - connecting the computer to their own WorkTrust account ("connect");
33
+ - the counter and the hooks it installs in their AI tools, and the reports
34
+ they send to their own account;
35
+ - importing their own history and data exports to WorkTrust ("history" and
36
+ export import); and
37
+ - anything else in the Software that talks to the WorkTrust service.
38
+
39
+ c. Without a WorkTrust account, beyond preserve, you may only install the
40
+ Software and run it to evaluate it.
41
+
42
+ You may make the copies needed to install and run the Software and a reasonable
43
+ number of backup copies. You may read the source code to understand and check
44
+ what the Software does on your computer, and you may test it on your own computer
45
+ to report a security problem as described in SECURITY.md. Being able to read the
46
+ source code gives you no other right.
47
+
48
+ 3. What you may not do
49
+
50
+ Except as section 2 allows or mandatory law permits, you may not:
51
+
52
+ a. copy the Software or any part of it;
53
+ b. modify, adapt, translate or create derivative works of the Software or any
54
+ part of it;
55
+ c. decompile, disassemble or reverse-engineer the Software;
56
+ d. sublicense, sell, rent, lend, redistribute, publish or otherwise make the
57
+ Software or any part of it available to anyone else;
58
+ e. use the Software, its code or any part of it in another product or service,
59
+ or to build, train or improve a product or service that competes with
60
+ WorkTrust;
61
+ f. remove or change any copyright notice, this licence or any other notice in
62
+ the Software; or
63
+ g. present a modified or rebuilt copy as the Software or as an official
64
+ WorkTrust release.
65
+
66
+ Nothing in this licence restricts what mandatory law allows a lawful user of a
67
+ computer program to do, including the acts permitted by Articles 5 and 6 of
68
+ Directive 2009/24/EC as implemented in the Dutch Copyright Act (Auteurswet).
69
+
70
+ 4. Ownership and all rights reserved
71
+
72
+ The Software is licensed, not sold. Oravtia B.V. (WorkTrust) retains all right,
73
+ title and interest in the Software and keeps all rights that this licence does
74
+ not expressly grant. This licence grants no right to the name WorkTrust, its
75
+ logos or other marks.
76
+
77
+ The archives, reports and other files the Software produces from your own data
78
+ belong to you. This licence gives Oravtia B.V. no right to them; what you send to
79
+ the WorkTrust service is governed by the service's terms and privacy statement.
80
+
81
+ 5. Earlier versions
82
+
83
+ Versions of the Software up to and including 0.10.17 were published under the MIT
84
+ licence. Those versions remain available under that licence for those versions
85
+ only. This licence applies to version 0.10.18 and every later version.
86
+
87
+ 6. Third-party components
88
+
89
+ The npm package contains no third-party code and has no dependencies.
90
+
91
+ The PyPI package's platform wheels include a Node.js runtime (the "node" or
92
+ "node.exe" executable) under the Node.js licence, shipped beside it as
93
+ worktrust_cli/runtime/LICENSE. The WorkTrust preservation kit includes the
94
+ WorkTrust AI evidence agent kit in its "agent" folder, under the licence shipped
95
+ as agent/LICENSE. Those components remain under their own licences, and this
96
+ licence does not limit the rights those licences give you.
97
+
98
+ 7. The WorkTrust service
99
+
100
+ Using the WorkTrust service is governed by its own terms at
101
+ https://worktrust.io/eula and its privacy statement at
102
+ https://worktrust.io/privacy. This licence grants no right to the service.
103
+
104
+ 8. Termination
105
+
106
+ This licence ends automatically if you breach it. Oravtia B.V. may also end it by
107
+ notice, for example when it ends the WorkTrust service. When it ends, you must
108
+ stop using the Software and delete every copy you hold. Sections 3, 4, 9, 10 and
109
+ 11 survive termination.
110
+
111
+ 9. No warranty
112
+
113
+ The Software is provided "as is" and "as available", without warranty of any
114
+ kind, express or implied, including any warranty of merchantability, fitness for
115
+ a particular purpose, non-infringement, or that it is free of errors or will run
116
+ without interruption.
117
+
118
+ 10. Limitation of liability
119
+
120
+ To the fullest extent the law allows, Oravtia B.V. is not liable for any damage
121
+ arising from or in connection with the Software or its use, whether in contract,
122
+ tort or otherwise, including loss of data, profit or business. This does not
123
+ limit liability for intent or deliberate recklessness, or any liability that
124
+ cannot be limited under mandatory law, including consumer protection law.
125
+
126
+ 11. Governing law
127
+
128
+ This licence is governed by the laws of the Netherlands. Disputes are submitted
129
+ to the competent court in the Netherlands, without limiting any mandatory right
130
+ a consumer has to bring proceedings in, or rely on the law of, their own country.
131
+
132
+ Contact: hello@worktrust.io
package/README.md CHANGED
@@ -306,6 +306,12 @@ sends a manifest and one zip per category (0.10.17): give a `conversations-000.z
306
306
  conversations parts lying beside it; its download links are never followed. The other parts (memories, projects,
307
307
  artifacts, frames, design chats, the light metadata) are refused by name and never opened.
308
308
 
309
+ Any size (0.10.18): the file is read as a stream, a conversation at a time, so a gigabyte export reads in a bounded
310
+ memory; it keeps of each conversation only the clocks, roles, ids and a model name it needs (the counter, the text it
311
+ reads here). Bounds: a file to 64 GB, two million conversations, nesting to 64; a single conversation whose kept
312
+ fields pass 8 MB is left out and counted, never the export. A zip over 4 GB (zip64) is refused by name: unpack it and
313
+ give `conversations.json`. Every 64 MB a line on stderr says how many megabytes and conversations were read, nothing else.
314
+
309
315
  - **What goes.** Per run of activity in a conversation (0.10.17; before, one line per conversation and day): a run
310
316
  ends where two messages lie more than thirty minutes apart, the gap after which the hook calls a stretch finished, so a
311
317
  conversation left open for hours is two windows, not one. Cutting changes the windows and never the total: the five
@@ -323,14 +329,17 @@ artifacts, frames, design chats, the light metadata) are refused by name and nev
323
329
  one question; Enter is No; `--dry-run` shows the plan and stops. On a coupled computer the bare `npx worktrust` looks
324
330
  in your Downloads folder (its top level only) for such an export it has not sent, offers it the same way, once, and
325
331
  remembers your answer by the file's digest, never by a conversation.
326
- - **The export replaces.** On the days the export covers, its lines count and the earlier lines of that product a
327
- connector brought in (an AI signed in on the web, which selects what it reports) are withdrawn and kept, each with a
328
- correction row, so a day counts once; the command says how many. A connector line that arrives later for a day the
329
- export covers is held. Lines an AI app reported earlier under no product may describe the same conversations: the
330
- command says how many lie on those days and asks once, `Also withdraw N earlier lines an AI app reported for these
331
- days?`; Enter is No, `--yes` keeps No, `--dry-run` sends nothing, and only a typed yes or `--withdraw-unknown` said
332
- in advance withdraws exactly those lines on exactly those days, the same way, never deleting. Never touched: another
333
- product, your own measured sessions, a private line, a day the export does not cover.
332
+ - **The export replaces.** On the days the export covers, its lines count and the earlier lines of that product your
333
+ connector of the same vendor reported (an AI signed in on the web, which selects what it reports) are withdrawn and
334
+ kept, each with a correction row, so a day counts once; the command says how many. A connector line that arrives later
335
+ for a day the export covers is held. Lines that connector reported earlier under no product may describe the same
336
+ conversations: the command says how many lie on those days and asks once, `Also withdraw N earlier lines your Claude
337
+ connector reported for these days?`; Enter is No, `--yes` keeps No, `--dry-run` sends nothing, and only a typed yes or
338
+ `--withdraw-unknown` said in advance withdraws exactly those lines on exactly those days, the same way, never deleting.
339
+ The count in the question is the door's own, and it is the count the yes withdraws. Never touched, whatever you answer
340
+ (0.10.19, held on the server): any line a computer of yours wrote (measured sessions, a rebuilt or archived history, an
341
+ earlier export send), a measured line, a connector of another vendor, another product, a private line, a day the export
342
+ does not cover.
334
343
 
335
344
  ## A copied session folder, and the local archive key
336
345
 
@@ -381,9 +390,17 @@ which would delete every password on the computer, and many people rightly keep
381
390
  | **Reads** | Which AI apps are installed (the settings folders of Claude Code, Codex, Cursor, Gemini CLI, Antigravity, VS Code, Windsurf, Hermes Agent, Goose, OpenCode, OpenClaw), and the chats Cursor and Copilot keep (clocks, models and token counts only). Afterwards, the session hook reads Claude Code's, Codex's and Antigravity's session files, the session databases of Hermes, Goose, OpenCode and OpenClaw, and the chats Cursor and Copilot keep, on this computer to measure durations and token counts. |
382
391
  | **Writes** | The key file (above). In each AI app's MCP settings (`~/.claude.json` through `claude mcp`, `~/.cursor/mcp.json`, `~/.codex/config.toml`, `~/.gemini/settings.json`, `~/.gemini/config/mcp_config.json` (Antigravity), VS Code's `mcp.json`, Windsurf's `mcp_config.json`, Hermes's and Goose's `config.yaml`, OpenCode's `opencode.json`, OpenClaw through `openclaw mcp set`) a WorkTrust entry that runs the local bridge, with no key in it; three hooks in `~/.claude/settings.json`; the bridge, the hook and the counter in `~/.worktrust/`. |
383
392
  | **Sends, to pair** | This computer's system (macOS, Windows, Linux) and host name. WorkTrust also records the network address the request arrives from, as a security record kept 90 days and seen only by WorkTrust staff (to answer a theft or fraud report). |
384
- | **Sends, afterwards** | Per measured stretch of work, and nothing else: its durations (`seconds`, `duration_basis`, `model_seconds`, `tool_seconds`, `human_seconds`, `idle_seconds`), token counts (`tokens_in`, `tokens_out`, `tokens_cache_read`, `tokens_cache_write`, `token_basis`), the model's name (`model`), the kind of work and one layer keyword such as frontend or backend (`title`, `kind`, `layer`, `layers`), closed technology keys (`stack`) and separate display-only tool keys (`tools_seen`), counts of how you work (`exchanges`, `turn_basis`, `agent_runs`, `agent_seconds`, `agent_peak`, `interrupts`, `steers`), when it began and ended and your computer's offset from UTC (`started_at`, `at`, `utc_offset`), what the CLI derived from the stretch as kinds and counts (`evidence`: checks run and failed, delivery steps and whether a check passed first, recoveries, delegation, context, routing, steering, refusals, planning, files and test files changed, the complexity class; with the WorkTrust mod on, its `live` counts from inside Claude Code: permission answers, effort levels, context and plan-limit peaks, subagents and setup; never a word, a command or a path; since 0.8.3, `live` since 0.10.15), a hash of the session and the day (`stretch_ref`), `steered_from` when it ran over SSH, and, so the work can be matched to GitHub, the repository as its remote's owner/name (`repo`, only for a repository on GitHub, GitLab or Bitbucket, the hosts WorkTrust can couple; never one on your own or a client's server) and the hashes of the commits made in the stretch (`commits`, never their messages). |
393
+ | **Sends, afterwards** | Per measured stretch of work, and nothing else: its timeline and basis (`timeline`, `duration_basis`: since 0.10.20 the CLI sends no duration and no hours in layers of its own; WorkTrust computes them from the timeline), token counts (`tokens_in`, `tokens_out`, `tokens_cache_read`, `tokens_cache_write`, `token_basis`), the model's name (`model`), the kind of work and one layer keyword such as frontend or backend (`title`, `kind`, `layer`, `layers`), closed technology keys (`stack`) and separate display-only tool keys (`tools_seen`), how the turns are counted (`turn_basis`; the counts themselves WorkTrust reads from the timeline; the subagents' seconds wait for a timeline per run), when it began and ended and your computer's offset from UTC (`started_at`, `at`, `utc_offset`), what the CLI derived from the stretch as kinds and counts (`evidence`: checks run and failed, delivery steps and whether a check passed first, recoveries, delegation, context, routing, steering, refusals, planning, files and test files changed, the complexity class; with the WorkTrust mod on, its `live` counts from inside Claude Code: permission answers, effort levels, context and plan-limit peaks, subagents and setup; never a word, a command or a path; since 0.8.3, `live` since 0.10.15), a hash of the session and the day (`stretch_ref`), `steered_from` when it ran over SSH, the stretch's clock as step kinds and whole seconds so WorkTrust computes the figures itself (`timeline`, since 0.10.20: never a word, a tool name or a path), the CLI's version and a digest of its own files as signals (`cli_version`, `cli_digest`), and, so the work can be matched to GitHub, the repository as its remote's owner/name (`repo`, only for a repository on GitHub, GitLab or Bitbucket, the hosts WorkTrust can couple; never one on your own or a client's server) and the hashes of the commits made in the stretch (`commits`, never their messages). |
385
394
  | **Never sends** | Prompts, answers, code, file names or paths, commit messages, branch names, or the name of your local folder (a repository is named only as its remote's owner/name, above). The door has no field for them and refuses a submission that carries them. |
386
395
 
396
+ **The clock is WorkTrust's, not this computer's (0.10.20).** A measured stretch travels as its timeline: one step per
397
+ transcript line, as a kind (your turn, the model's answer, a tool call, a tool's result, a question that waited for you, an
398
+ interruption) and the whole seconds since the previous step. Nothing of a message is in it: no word, no tool name, no path.
399
+ This CLI computes no duration and no hours of its own and carries no rule for doing so; WorkTrust computes the seconds, the
400
+ hours in layers and the turns from the timeline when it arrives, and the figures the terminal shows after a send are
401
+ WorkTrust's receipt. The line also names the CLI's version and a digest of its own files, as signals of which code ran,
402
+ never as proof.
403
+
387
404
  ## Why you can check it
388
405
 
389
406
  - **No dependencies and no install scripts.** Plain files, readable
@@ -437,5 +454,8 @@ Security reports: see [SECURITY.md](SECURITY.md).
437
454
 
438
455
  ## Licence
439
456
 
440
- MIT, for this command-line tool only (see LICENSE). The WorkTrust service it couples to is not
441
- open source.
457
+ Use is allowed under the WorkTrust CLI Licence; copying, modifying or redistributing the code is not. See LICENSE.
458
+ The licence is free of charge. `preserve` (with `--apply`, `--archive`, `--verify` and `--summary`) is for everyone,
459
+ with or without an account; every other feature is for WorkTrust account holders, and without an account the rest may
460
+ only be installed and evaluated. The archives and files the CLI makes from your own data are yours.
461
+ From 0.10.18 on; versions up to and including 0.10.17 were published under the MIT licence and stay under it.
package/SECURITY.md CHANGED
@@ -59,10 +59,14 @@ potentially identifying metadata; see the README's complete payload list.
59
59
 
60
60
  Agent-shell detection is a workflow safeguard, not protection from malicious same-user software.
61
61
  Agents must follow AGENTS.md. From 0.10.12 the bare invocation is covered by the same guard as explicit
62
- person-only commands; help and version are safe informational paths. Export parsing limits ZIP size
63
- to 2 GiB, directory size to 8 MiB, compressed/decoded JSON to 64 MiB, nesting to 64, structural tokens
64
- to 1,000,000 and conversations to 50,000. Larger inputs are refused. These are resource bounds,
65
- not a sandbox or independent validation of the export's truth.
62
+ person-only commands; help and version are safe informational paths. An export is read as a stream
63
+ (0.10.18) through a JSON walk of our own that keeps a fixed set of fields per conversation and skips the
64
+ rest unbuilt; a zip entry is inflated as a stream. Bounds: a file (bare or zip) to 64 GiB, a zip directory
65
+ to 8 MiB, nesting to 64, conversations to 2,000,000, a conversation's kept fields to 8 MiB (the counter's
66
+ text, read locally, to 64 MiB; a conversation past its budget is left out and counted), an export held whole
67
+ to 512 MiB of kept fields; zip64 is refused. Peak memory is bounded by the chunk in flight and one
68
+ conversation's kept fields (`scripts/test-export-memory.mjs` holds it under 200 MiB for a gigabyte). These
69
+ are resource bounds, not a sandbox or independent validation of the export's truth.
66
70
 
67
71
  ## Verifying a version
68
72