diff --git a/CHANGELOG.md b/CHANGELOG.md
index 886c25432..db82f4739 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,3 +1,44 @@
+### 2.12.0 (Unreleased)
+### Features/Bug Fixes
+* fix(hooks): fail closed on unmodeled payload analysis (#573)
+* perf(security): cache pure security-view predicates (#570)
+* perf(security): accelerate token-gap classification with ASCII fast paths (#569)
+* fix(input): download raw content for GitHub and GitLab blob URLs (#566)
+* fix(cli): emit recursive SARIF reports to stdout (#565)
+* fix(security): keep line breaks out of letter-spacing runs (#564)
+* feat(cli): add streamed progress, file discovery, and Gemini 3.5 Flash metadata (#7)
+* fix(providers): honor model registry overrides for CLI providers (#463)
+* test(input): inject secure-open failures on the primitive the platform uses (#503)
+* feat(analyzer): add OpenCode-native SkillSpector invocation skill and tool (#537)
+* fix(input): expand 8.3 short DOS names before the opened-handle check (#484)
+* test(build-context): write fixture content with the newlines the assertions pin (#505)
+* test(build-context): guard the second FIFO case on mkfifo availability (#502)
+* test(build-context): read the nested OMS fixture without newline translation (#518)
+* test(yara): write the non-ASCII malformed-rule fixture as UTF-8 (#504)
+* fix(security): account for excluded executables (#548)
+* feat(providers): add opencode_cli semantic-scan provider (#536)
+* fix(llm): retarget the deadline on the existing client instead of rebuilding it (#520)
+* fix: preserve Markdown context during JSON command recovery (#559)
+* fix(batch-scan): reuse validated cache for multilingual analysis (#558)
+* fix(as3): ignore current-skill self references (#506)
+* fix: classify companion CLI documentation by context (#547)
+* fix(security): enforce discovery and requested analysis completeness (#410)
+* fix(security): add bounded reconstruction for letter-spaced P3/P4 instructions with an AE6 fallback for some ambiguous reconstructions (#470)
+* feat(cli): add opt-in --fail-on-findings across single-skill, recursive, and MCP registry scans (#469)
+* fix(security): preserve exact finding identity, evidence, and occurrence locations (#409)
+* fix: report oversized-file coverage gaps as AE7 and bound their LLM input (#509)
+* fix: avoid false analysis limits in documentation (#516)
+* fix: scan JSON quote candidates in linear time (#521)
+* feat: configure the static analysis time allowance (#522)
+* fix(cli): emit recursive JSON reports to stdout (#467)
+* fix(analyzer): project manifest version for RP3 (#474)
+* fix(supply-chain): prefer exact known-package matches in SC6 (#530)
+* Fix false local references from slash-separated prose (#451)
+* test(build-context): skip symlink cases where the platform refuses symlinks (#501)
+* docs(providers): align setup guidance (#434)
+* docs(readme): add HVTrust badge (#428)
+* docs(readme): clarify research background counts (#543)
+---
 ### 2.11.2 (Thursday, September 10, 2026)
 ### Features/Bug Fixes
 * fix: prevent duplicate reference coverage from causing fatal scan-accounting errors (#507)
diff --git a/docs/release/skillspector-2.12.0.md b/docs/release/skillspector-2.12.0.md
new file mode 100644
index 000000000..7fcdbaf50
--- /dev/null
+++ b/docs/release/skillspector-2.12.0.md
@@ -0,0 +1,147 @@
+# SkillSpector v2.12.0
+
+Release status: candidate; publication pending.
+
+## Summary
+
+SkillSpector 2.12.0 adds an opt-in CLI gate for any active finding, a configurable static-analysis allowance, an OpenCode CLI semantic-analysis provider, an OpenCode-native scan command and tool, Gemini 3.5 Flash registry guidance, and bounded interactive CLI progress with discovered-file visibility. It also fixes false AE1 incomplete-analysis results caused by ordinary Markdown and JSON documentation, makes discovery and requested-analysis gaps explicit, inventories excluded executable content, fails closed on unmodeled active-hook payloads, preserves distinct findings and their source locations, emits valid recursive SARIF to standard output, handles forge `/blob/` links as raw files, and reduces both false positives and repeated security-view work. Oversized files now produce an explicit coverage finding and bounded LLM input.
+
+## Highlights
+
+- Run semantic analysis through the new `opencode_cli` provider and invoke SkillSpector from OpenCode through a native `/skillspector` command and `skillspector_scan` tool.
+- Use `gemini-3.5-flash` through the existing OpenAI-compatible provider, and see bounded live progress plus a control-safe discovered-file tree during non-verbose interactive scans.
+- Emit merged recursive SARIF on standard output without status-text contamination, and scan GitHub/GitLab `/blob/` links as their raw file contents instead of forge HTML.
+- Inventory executable and loadable content in normally excluded locations, preserving coverage evidence and failing closed when referenced or otherwise outside inspection coverage.
+- Mark active hooks with unmodeled payload data flow as partial and opaque, retaining BH1 while preventing a misleading `SAFE` recommendation.
+- Reduce security-view CPU and tail latency with ASCII token-gap fast paths and bounded memoization while preserving findings and coverage outcomes.
+- Keep multilingual batch language detection and gap-fill on the graph's validated provider-eligible cache instead of rereading paths after inspection.
+- Retarget per-call LLM deadlines on existing clients to avoid connection-pool churn and closed-event-loop cleanup failures.
+- Honor model-registry overrides for CLI providers and accept valid Windows 8.3 aliases without weakening opened-handle validation.
+- Recognize complete JSON strings and Markdown code spans in their document context, avoiding false analysis limits while retaining analysis of their contents.
+- Keep delimiter pairing within Markdown blocks and table cells so unrelated documentation cannot hide unresolved runtime commands.
+- Scan JSON quote candidates in linear time and retain cancellation handling.
+- Preserve distinct full-evidence and rule identities, including findings with identical shortened previews, and retain precise locations for repeated occurrences.
+- Keep requested but unavailable or incomplete semantic analysis visible in completeness metadata and strict CLI/MCP decisions.
+- Classify selected OAuth, signed self-update, and warned installer documentation based on surrounding context, subject to the PE3 and RA1 gaps recorded under Known Limitations.
+- Suppress AS3 only when a literal `skills/<name>/SKILL.md` path identifies the skill currently being scanned, while retaining peer-skill, transformed, and enumeration findings.
+
+## Added
+
+- `SKILLSPECTOR_PROVIDER=opencode_cli` runs semantic analysis through a local OpenCode login. The verified deny-all policy requires exactly OpenCode 1.18.30; authentication, version, policy, empty-output, or event-envelope failures fail closed. `SKILLSPECTOR_MODEL` remains optional ([#536](https://github.com/NVIDIA/SkillSpector/pull/536)).
+- The repository-provided OpenCode extension adds a static-by-default `/skillspector` command and `skillspector_scan` tool. Copy `.opencode/` from a checkout to install it; the wheel does not install the extension. It resolves the binary from `SKILLSPECTOR_BIN`, a worktree `.venv`, or `PATH`, requests required host capabilities before launch, rejects symlinked target/output/binary paths, and uses a 120-second timeout with bounded, redacted output. Semantic analysis is opt-in through `noLlm=false` and the provider environment ([#537](https://github.com/NVIDIA/SkillSpector/pull/537)).
+- `SKILLSPECTOR_MAX_STATIC_ANALYSIS_SECONDS_PER_ARTIFACT` configures the static pattern and YARA time allowance per artifact; its default increases from 30 to 300 seconds. The remaining workflow deadline still bounds both analyzers.
+- `skillspector scan --fail-on-findings` exits with code 1 when a scan reports any active finding, including findings below the default risk-score threshold. It applies to single-skill, recursive, and MCP registry scans. Skill scans evaluate active findings after suppression ([#469](https://github.com/NVIDIA/SkillSpector/pull/469)).
+- `gemini-3.5-flash` model budgets are registered for the root and OpenAI-compatible registries, with a Gemini setup example using the existing `openai` provider ([#7](https://github.com/NVIDIA/SkillSpector/pull/7)).
+
+## Changed
+
+- Non-verbose scans attached to an interactive terminal now stream graph execution and render bounded progress, completed analyzer rules, and a control-safe discovered-file tree to standard error. Machine-readable standard output and exit-code behavior remain intact ([#7](https://github.com/NVIDIA/SkillSpector/pull/7)).
+- Printable-ASCII token-gap scans use precomputed classification tables and a whole-text fast path; repeated pure security-view predicates use bounded memoization. The measured 901-skill corpus retained byte-identical findings and coverage outcomes while reducing p95, p99, and total CPU cost ([#569](https://github.com/NVIDIA/SkillSpector/pull/569), [#570](https://github.com/NVIDIA/SkillSpector/pull/570)).
+- CLI-backed providers now honor `SKILLSPECTOR_MODEL_REGISTRY` for context and output-token limits. Missing entries retain the existing fallback behavior; malformed registry structures or invalid and non-positive budgets warn and fall back ([#463](https://github.com/NVIDIA/SkillSpector/pull/463)).
+- The repository's `contrib/batch_scan` tool reuses the graph's validated `llm_file_cache` for multilingual language detection and gap-fill, avoiding a second raw filesystem read and retaining provider/local-only boundaries. `contrib` remains outside the wheel ([#558](https://github.com/NVIDIA/SkillSpector/pull/558)).
+- Align provider setup guidance, add the HVTrust badge, and update research background counts ([#434](https://github.com/NVIDIA/SkillSpector/pull/434), [#428](https://github.com/NVIDIA/SkillSpector/pull/428), [#543](https://github.com/NVIDIA/SkillSpector/pull/543)).
+- Reports retain requested LLM intent separately from runtime availability. Incomplete semantic execution remains visible through aggregate reports and CLI/MCP installation gates ([#410](https://github.com/NVIDIA/SkillSpector/pull/410)).
+
+## Fixed
+
+- Active hook declarations whose payload data flow remains unmodeled now retain the BH1 mechanism finding and record partial `opaque_content` coverage, making completeness false and preventing a `SAFE` recommendation ([#573](https://github.com/NVIDIA/SkillSpector/pull/573)).
+- Letter-spacing reconstruction no longer treats a logical line break as the start of a token run, so CRLF or punctuation on the previous heading cannot turn identifier-adjacent text into false P3/P4 findings; AE6 still records the ambiguous form ([#564](https://github.com/NVIDIA/SkillSpector/pull/564)).
+- Recursive `--format sarif` scans without `--output` now emit the merged SARIF log to standard output. SARIF advisories, progress, verbose status, and transitive warnings stay on standard error so the output remains parseable ([#565](https://github.com/NVIDIA/SkillSpector/pull/565)).
+- GitHub and GitLab `/blob/` file URLs are rewritten to their raw-file forms before download, preventing scans of forge HTML in place of the requested file ([#566](https://github.com/NVIDIA/SkillSpector/pull/566)).
+- Retarget dynamic workflow deadlines on existing OpenAI, Anthropic, and agent-CLI clients instead of constructing a new client per call, preventing connection-pool churn and closed-event-loop cleanup errors while preserving retry and concurrency behavior ([#520](https://github.com/NVIDIA/SkillSpector/pull/520)).
+- Preserve whole-document Markdown ownership when recovering commands from validated JSON strings so fenced or literal content is not reinterpreted as standalone Markdown and unresolved commands cannot become a clean result through unrelated JSON adjacency ([#559](https://github.com/NVIDIA/SkillSpector/pull/559)).
+- Inventory normally excluded executable and loadable content, retain root-coverage evidence, resolve explicit extensionless command paths, and inspect excluded ZIP-family containers within existing limits. Referenced executable exclusions or incomplete excluded-artifact inspection now emit HIGH SC9/incomplete evidence with the existing minimum score of 51 (`DO_NOT_INSTALL`); the narrow exception for direct, non-binary `.git/hooks/*.sample` files remains ([#548](https://github.com/NVIDIA/SkillSpector/pull/548)).
+- Expand Windows 8.3 short-name components before comparing a requested path with its opened handle, allowing valid paths under spaced profile directories while retaining reparse-point and fail-closed checks ([#484](https://github.com/NVIDIA/SkillSpector/pull/484)).
+- Avoid false AE1 results from valid JSON placeholders, inline code, list and blockquote containers, indented JSON, Markdown tables, and literal Make syntax ([#516](https://github.com/NVIDIA/SkillSpector/pull/516)).
+- Bound JSON quote traversal without repeatedly scanning overlapping suffixes ([#521](https://github.com/NVIDIA/SkillSpector/pull/521)).
+- Emit a HIGH AE7 analysis-evasion finding for per-file size limits that leave an artifact partially inspected, unless AE1 already covers that path. Supply a bounded text prefix and an explicit unreviewed-region marker to enabled LLM analysis; the unread region remains incomplete ([#509](https://github.com/NVIDIA/SkillSpector/pull/509)).
+- Preserve full-evidence fingerprints, concrete YARA rule identity, and precise occurrence locations through projection, deduplication, and report compaction. Separate findings are retained while duplicate projections of the same occurrence are collapsed ([#409](https://github.com/NVIDIA/SkillSpector/pull/409)).
+- Add bounded reconstruction for letter-spaced P3/P4 prompt instructions, retain source evidence, and use AE6 as an incomplete-coverage fallback for some ambiguous reconstructions. Some alternating-width short runs can evade both P3/P4 and AE6 (see Known Limitations) ([#470](https://github.com/NVIDIA/SkillSpector/pull/470)).
+- Discover skills inside dot-prefixed directories, retain inherited local-only restrictions for child skills, and partition transitive scan caching by those privacy restrictions ([#410](https://github.com/NVIDIA/SkillSpector/pull/410)).
+- Preserve incomplete discovery and requested semantic-analysis failures, including unavailable providers and mixed success/failure telemetry, instead of allowing a complete scan result ([#410](https://github.com/NVIDIA/SkillSpector/pull/410)).
+- Emit recursive JSON reports to standard output when no output path is provided ([#467](https://github.com/NVIDIA/SkillSpector/pull/467)).
+- Use the project manifest version for RP3 analysis ([#474](https://github.com/NVIDIA/SkillSpector/pull/474)).
+- Prefer exact known-package matches when evaluating SC6 package-name similarity ([#530](https://github.com/NVIDIA/SkillSpector/pull/530)).
+- Avoid treating slash-separated prose as local file references ([#451](https://github.com/NVIDIA/SkillSpector/pull/451)).
+- Reduce false-positive severity for companion CLI OAuth results and signed self-update documentation classified as benign, and provide contextual explanations for warned pipe-to-shell installers. The contextual classifiers have unresolved PE3 and RA1 fail-open paths (see Known Limitations) ([#547](https://github.com/NVIDIA/SkillSpector/pull/547)).
+- Ignore literal AS3 references to the current skill, derived from the scan-root basename or manifest name, without suppressing peer-skill or obfuscated-path access ([#506](https://github.com/NVIDIA/SkillSpector/pull/506)).
+- Skip symlink test cases when the platform refuses symlink creation ([#501](https://github.com/NVIDIA/SkillSpector/pull/501)).
+
+## Testing and Portability
+
+- Make secure-open, FIFO, newline-sensitive build-context, nested OMS, and non-ASCII YARA fixtures deterministic across Windows and non-POSIX environments ([#503](https://github.com/NVIDIA/SkillSpector/pull/503), [#502](https://github.com/NVIDIA/SkillSpector/pull/502), [#505](https://github.com/NVIDIA/SkillSpector/pull/505), [#518](https://github.com/NVIDIA/SkillSpector/pull/518), [#504](https://github.com/NVIDIA/SkillSpector/pull/504)).
+- Add dependency-free Node 22+ tests for the OpenCode tool helpers and an exact-head OpenCode TypeScript CI job ([#537](https://github.com/NVIDIA/SkillSpector/pull/537)).
+
+## Security
+
+- Active hook payloads whose data flow is not modeled remain visible through BH1 and now make analysis partial with `opaque_content`, preventing a complete or `SAFE` result for those hooks.
+- Excluded executable or loadable content is inventoried before exclusion. Referenced or out-of-coverage bytes now produce SC9 and incomplete-analysis evidence that blocks strict installation gates; the narrow exception is limited to direct, non-binary `.git/hooks/*.sample` files.
+- Multilingual batch analysis consumes only the validated provider-eligible snapshot, so language detection and gap-fill do not reread local-only or replaced path content after the core scan.
+- OpenCode integration is static by default. The native tool redacts common secret forms and bounds output, while the semantic provider uses argv/stdin and treats missing authentication, empty output, or unsupported event streams as failures.
+- Genuine removal instructions remain reportable. The covered unresolved-runtime-command controls retain incomplete coverage and fail strict CLI/MCP installation gates, including when semantic analysis succeeds. Additional runtime-selected command variants remain under investigation (see Known Limitations).
+- JSON string ownership preserves source evidence and does not exempt string contents from analysis.
+- Findings and exit status can change after upgrading: oversized files can add HIGH AE7 findings, letter-spaced instructions can produce P3/P4 or AE6 findings, and previously collapsed distinct matches can increase the retained finding count and risk score. Missing requested analysis remains incomplete even when static analysis finishes.
+- Context-aware companion CLI classification can lower severity, scores, or recommendations for documentation classified as benign. Coverage is not fail-closed for every token-transfer or self-update phrasing: the known PE3 and RA1 exceptions below can be incorrectly downgraded or missed. Warned pipe-to-shell installer findings remain reportable.
+- Literal current-skill references no longer produce AS3 findings. Peer-skill references, transformed or obfuscated paths, explicit enumeration, AS1, and AS2 remain reportable.
+
+## Breaking Changes and Migration
+
+- Interactive non-verbose scans now show progress and discovered files on standard error. Machine-readable JSON and SARIF remain clean on standard output; use `--verbose` to retain the non-streamed diagnostic path.
+- Scanning GitHub or GitLab `/blob/` links now analyzes raw file bytes rather than the forge viewer page. Findings and recommendations can change because the intended content is finally scanned.
+- Active hooks with unmodeled payload flows now produce incomplete coverage and cannot remain `SAFE`; identifier-adjacent letter spacing can lose false P3/P4 findings while retaining AE6.
+- `opencode_cli` is opt-in and requires an authenticated OpenCode 1.18.30 executable; other versions fail closed because their deny-all policy has not been verified. The OpenCode-native tool is installed by copying `.opencode/` from a checkout and defaults to static analysis; set `noLlm=false` and configure the provider environment to request semantic analysis.
+- Existing CLI-provider deployments that set `SKILLSPECTOR_MODEL_REGISTRY` now use its valid token budgets. Invalid or non-positive values warn and fall back instead of aborting.
+- Scans that previously treated excluded executable content as clean can now become incomplete with SC9 and `DO_NOT_INSTALL`; consumers should retain completeness and exclusion evidence.
+- No new configuration is required. Static analysis can now run longer within the existing workflow deadline. Set `SKILLSPECTOR_MAX_STATIC_ANALYSIS_SECONDS_PER_ARTIFACT=30` to retain the previous per-artifact allowance, and restart the SkillSpector process after changing the setting ([#522](https://github.com/NVIDIA/SkillSpector/pull/522)).
+- `--fail-on-findings` is opt-in. Combine it with `--fail-on-incomplete` when CI must reject either active findings or incomplete coverage. Code 2 still denotes an input or execution error; a nonzero exit alone does not identify which condition occurred.
+- Consumers should retain completeness/degradation metadata and inspect findings as well as exit status. Explicitly use `--no-llm` for an intended static-only scan; requesting LLM analysis without an available provider is incomplete.
+- Third-party dependency versions are unchanged from 2.11.2.
+
+## Deprecations
+
+- None.
+
+## Validation
+
+The release candidate is synchronized through main commit `548e5e0afd25595ef039c27cdeb283413c282c71` and was validated locally on the refreshed release-catalog working tree with Python 3.12.13:
+
+- `uv lock --check --offline` passed with the locked dependency set.
+- `make test-ci` passed 5,744 tests, with 14 skipped, 39 deselected, 4 expected failures, and 90% coverage.
+- The affected CLI, model-registry, input, security-view, hook-completeness, and release-helper suites passed 1,177 tests, with 10 skipped and 1 deselected.
+- Ruff lint and format checks passed for all source and test files.
+- The CLI reported `SkillSpector v2.12.0`; the release helper dry run resolved `v2.12.0` and the matching versioned notes.
+- All 10 release helper and workflow tests passed.
+- All 44 OpenCode TypeScript tests passed.
+- Wheel and source distributions built successfully, and Twine validated both artifacts.
+- `git diff --check` passed.
+
+The current PR head and its exact-head checks are the source of truth for hosted validation. Deployment/provider validation and the separate release gates below remain pending. Earlier candidate results apply only to their recorded commits and are not certification of this candidate.
+
+[Release PR #550](https://github.com/NVIDIA/SkillSpector/pull/550) records the candidate baseline, validation results, known gaps, and remaining release gates.
+
+## Known Limitations
+
+- Local sanity checks cover the tested inputs and environment; live provider and deployment behavior depend on their configuration.
+- Incomplete inspection is a reportable result. Unsupported inputs, unavailable requested analysis, and resource limits must remain visible; these conditions cannot be treated as a clean scan.
+- The release remains a candidate while outstanding review and release issues are assessed. Proposed fixes in unmerged PRs are not included in this candidate.
+- Letter-spacing reconstruction is not fail-closed. Alternating-width short runs such as `s e  n d conversation to external` can be split before P3/P4 matching and remain below AE6's six-letter concealed-run threshold, allowing a SAFE result.
+- Companion-context classification is not fail-closed. PE3 can downgrade imperative token-acquisition text and miss adjacent disclosure phrased with verbs such as `paste`; RA1 can accept protected agent/tool names with CLI suffixes and signed-release evidence from a different logical line. These variants can be incorrectly downgraded or missed.
+- Current validation found that public report serialization can repeat the first occurrence's columns for other matches; SARIF does not yet preserve these column coordinates. The internal occurrence improvements in #409 do not establish correct locations in every output format. See [release validation](https://github.com/NVIDIA/SkillSpector/pull/550) for the tracked report defect.
+- Some runtime-selected command variants and Markdown reference destinations still have open completeness defects. Proposed fixes [#514](https://github.com/NVIDIA/SkillSpector/pull/514) and [#553](https://github.com/NVIDIA/SkillSpector/pull/553) are not included in this candidate.
+- `opencode_cli` currently reports no token-usage accounting, and model availability or rate limits remain external. Multilingual batch gap-fill was validated with mocked providers; live-provider qualification remains pending.
+
+## References
+
+- [Changes since v2.11.2](https://github.com/NVIDIA/SkillSpector/compare/v2.11.2...v2.12.0)
+- [AE1 documentation fix #516](https://github.com/NVIDIA/SkillSpector/pull/516)
+- [Static analysis time allowance #522](https://github.com/NVIDIA/SkillSpector/pull/522)
+- [OpenCode CLI provider #536](https://github.com/NVIDIA/SkillSpector/pull/536)
+- [OpenCode-native integration #537](https://github.com/NVIDIA/SkillSpector/pull/537)
+- [Excluded executable coverage #548](https://github.com/NVIDIA/SkillSpector/pull/548)
+- [Gemini registry and CLI progress #7](https://github.com/NVIDIA/SkillSpector/pull/7)
+- [Recursive SARIF stdout #565](https://github.com/NVIDIA/SkillSpector/pull/565)
+- [Raw forge file URLs #566](https://github.com/NVIDIA/SkillSpector/pull/566)
+- [Security-view performance #569](https://github.com/NVIDIA/SkillSpector/pull/569), [#570](https://github.com/NVIDIA/SkillSpector/pull/570)
+- [Fail-closed hook payload coverage #573](https://github.com/NVIDIA/SkillSpector/pull/573)
+
+Prepared by Codex for Mohit Gupta.
diff --git a/pyproject.toml b/pyproject.toml
index 5b7fa7b29..e04a8ccaa 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "skillspector"
-version = "2.11.2"
+version = "2.12.0"
 description = "SkillSpector: Security scanner for AI agent skills (Claude Code, Cursor, and similar). Scans skills for vulnerabilities, malicious patterns, and security risks before installation. Supports Git repos, URLs, zips, and local directories; runs static pattern checks and optional LLM semantic analysis; outputs terminal, JSON, and Markdown reports with risk scoring."
 readme = "README.md"
 license = "Apache-2.0"
diff --git a/uv.lock b/uv.lock
index 3a0df7ea1..c2dd47c50 100644
--- a/uv.lock
+++ b/uv.lock
@@ -2687,7 +2687,7 @@ wheels = [
 
 [[package]]
 name = "skillspector"
-version = "2.11.2"
+version = "2.12.0"
 source = { editable = "." }
 dependencies = [
     { name = "boto3" },